News: Technology News

Samsung and AMD extend Exynos GPU partnership, hope to find customers

Thu, 06 Apr 2023 21:12:48 +0000

Samsung doesn't use Exynos in flagships anymore, but maybe mid-rangers will buy in.

Samsung Electronics and AMD are extending their smartphone chip agreement. In a press release today, the two companies said they "signed a multi-year agreement extension to bring multiple generations of high-performance, ultra-low-power AMD Radeon graphics solutions to an expanded portfolio of Samsung Exynos SoCs." Even Samsung is reluctant to use Samsung chips these days, so it's not clear what devices these AMD GPUs will land in.

Samsung's chip division and AMD have already done a generation of an Exynos SoC with an AMD GPU. That chip was the Exynos 2200, with its "Samsung Xclipse 920 GPU" that was co-developed by AMD. Samsung's phone division—which doesn't necessarily have a bias toward the chip division's products—shipped that chip in the S22 in some regions like Europe, while shipping the S22 with Qualcomm chips in other regions, like the US and China. Exynos chips have a bad reputation for constantly having lower performance compared to Qualcomm chips, and the Exynos 2200 was no exception. The chip didn't do well against the Snapdragon 8 Gen 1 in benchmarks and power usage, and Samsung fans once again had to deal with getting "inferior phones" depending on what country they lived in.

The one win the Xclipse GPU had over Qualcomm was in ray tracing. In today's four-paragraph press release, Samsung and AMD point out twice that "Xclipse was the industry’s first mobile GPU with hardware-accelerated ray tracing." Samsung's 2022 chip is better at ray tracing than Qualcomm's 2023 chip! The problem is, being an early adopter of ray tracing isn't really relevant for mobile gaming. Mobile games are built for a causal audience and target mass-market hardware. With the need to also balance battery life, that market doesn't value high graphics fidelity. As you can see with Lenovo killing off its gaming smartphone line, the attempts to bring hardcore gaming values to smartphones have not found a huge audience.

Samsung's chip division brands these Exynos chips in line with the phones from Samsung's phone division, but the two groups don't see eye to eye on the chip's market viability. While only a portion of the Galaxy S22s shipped with Exynos 2200 chips, the Galaxy S23 skipped the Exynos 2300 entirely and went all Qualcomm, in every region. The Exynos 2300 is rumored to still exist and might debut in some special "Fan editions" of Samsung's products, which always feel like they are built out of spare parts. Exynos and AMD are both clinging to smartphone relevance, and it's not clear if this partnership will have any customers.

The line about bringing Radeon graphics to "an expanded portfolio of Samsung Exynos SoCs" might get the two companies some design wins. Samsung's phone division still ships Exynos chips on mid- and low-end devices, so maybe that's where we'll start to see Radeon tech hit the market. It doesn't make a ton of sense to bring high-end desktop features to mid-range mobiles, but the two companies are probably desperate for any kind of market share.

Samsung and AMD extend Exynos GPU partnership, hope to find customers

Why ChatGPT and Bing Chat are so good at making things up

Thu, 06 Apr 2023 21:11:06 +0000

A look inside the hallucinating artificial minds of the famous text prediction bots.

Over the past few months, AI chatbots like ChatGPT have captured the world's attention due to their ability to converse in a human-like way on just about any subject. But they come with a serious drawback: They can present convincing false information easily, making them unreliable sources of factual information and potential sources of defamation.

Why do AI chatbots make things up, and will we ever be able to fully trust their output? We asked several experts and dug into how these AI models work to find the answers.

“Hallucinations”—a loaded term in AI

AI chatbots such as OpenAI's ChatGPT rely on a type of AI called a "large language model" (LLM) to generate their responses. An LLM is a computer program trained on millions of text sources that can read and generate "natural language" text—language as humans would naturally write or talk. Unfortunately, they can also make mistakes.

In academic literature, AI researchers often call these mistakes "hallucinations." But that label has grown controversial as the topic becomes mainstream because some people feel it anthropomorphizes AI models (suggesting they have human-like features) or gives them agency (suggesting they can make their own choices) in situations where that should not be implied. The creators of commercial LLMs may also use hallucinations as an excuse to blame the AI model for faulty outputs instead of taking responsibility for the outputs themselves.

Still, generative AI is so new that we need metaphors borrowed from existing ideas to explain these highly technical concepts to the broader public. In this vein, we feel the term "confabulation," although similarly imperfect, is a better metaphor than "hallucination." In human psychology, a "confabulation" occurs when someone's memory has a gap and the brain convincingly fills in the rest without intending to deceive others. ChatGPT does not work like the human brain, but the term "confabulation" arguably serves as a better metaphor because there's a creative gap-filling principle at work, as we'll explore below.

The confabulation problem

It's a big problem when an AI bot generates false information that can potentially mislead, misinform, or defame. Recently, The Washington Post reported on a law professor who discovered that ChatGPT had placed him on a list of legal scholars who had sexually harassed someone. But it never happened—ChatGPT made it up. The same day, Ars reported on an Australian mayor who allegedly found that ChatGPT claimed he had been convicted of bribery and sentenced to prison, a complete fabrication.

Shortly after ChatGPT's launch, people began proclaiming the end of the search engine. At the same time, though, many examples of ChatGPT's confabulations began to circulate on social media. The AI bot has invented books and studies that don't exist, publications that professors didn't write, fake academic papers, false legal citations, non-existent Linux system features, unreal retail mascots, and technical details that don't make sense.

And yet despite ChatGPT's predilection for casually fibbing, counter-intuitively, its resistance to confabulation is why we're even talking about it today. Some experts note that ChatGPT was technically an improvement over vanilla GPT-3 (its predecessor model) because it could refuse to answer some questions or let you know when its answers might not be accurate.

"A major factor in Chat's success is that it manages to suppress confabulation enough to make it unnoticeable for many common questions," said Riley Goodside, an expert in large language models who serves as staff prompt engineer at Scale AI. "Compared to its predecessors, ChatGPT is notably less prone to making things up."

If used as a brainstorming tool, ChatGPT's logical leaps and confabulations might lead to creative breakthroughs. But when used as a factual reference, ChatGPT could cause real harm, and OpenAI knows it.

Not long after the model's launch, OpenAI CEO Sam Altman tweeted, "ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness. It's a mistake to be relying on it for anything important right now. It’s a preview of progress; we have lots of work to do on robustness and truthfulness." In a later tweet, he wrote, "It does know a lot, but the danger is that it is confident and wrong a significant fraction of the time."

What's going on here?

How ChatGPT works

An AI-generated image of a chatbot hovering in the library, as one does.

Benj Edwards / Stable Diffusion

To understand how a GPT model like ChatGPT or Bing Chat confabulates, we have to know how GPT models work. While OpenAI hasn't released the technical details of ChatGPT, Bing Chat, or even GPT-4, we do have access to the research paper that introduced their precursor, GPT-3, in 2020.

Researchers build (train) large language models like GPT-3 and GPT-4 by using a process called "unsupervised learning," which means the data they use to train the model isn't specially annotated or labeled. During this process, the model is fed a large body of text (millions of books, websites, articles, poems, transcripts, and other sources) and repeatedly tries to predict the next word in every sequence of words. If the model's prediction is close to the actual next word, the neural network updates its parameters to reinforce the patterns that led to that prediction.

Conversely, if the prediction is incorrect, the model adjusts its parameters to improve its performance and tries again. This process of trial and error, though a technique called "backpropagation," allows the model to learn from its mistakes and gradually improve its predictions during the training process.

As a result, GPT learns statistical associations between words and related concepts in the data set. Some people, like OpenAI Chief Scientist Ilya Sutskever, think that GPT models go even further than that, building a sort of internal reality model so they can predict the next best token more accurately, but the idea is controversial. The exact details of how GPT models come up with the next token within their neural nets are still uncertain.

"what does it mean to predict the next token well enough? ... it means that you understand the underlying reality that led to the creation of that token"

excellent explanation by @ilyasut, and thoughts on the crucial question: how far can these systems extrapolate beyond human? pic.twitter.com/v8zFQWvxWY
— BioBootloader (@bio_bootloader) March 28, 2023

In the current wave of GPT models, this core training (now often called "pre-training") happens only once. After that, people can use the trained neural network in "inference mode," which lets users feed an input into the trained network and get a result. During inference, the input sequence for the GPT model is always provided by a human, and it's called a "prompt." The prompt determines the model's output, and altering the prompt even slightly can dramatically change what the model produces.

For example, if you prompt GPT-3 with "Mary had a," it usually completes the sentence with "little lamb." That's because there are probably thousands of examples of "Mary had a little lamb" in GPT-3's training data set, making it a sensible completion. But if you add more context in the prompt, such as "In the hospital, Mary had a," the result will change and return words like "baby" or "series of tests."

Here's where things get a little funny with ChatGPT, since it's framed as a conversation with an agent rather than just a straight text-completion job. In the case of ChatGPT, the input prompt is the entire conversation you've been having with ChatGPT, starting with your first question or statement and including any specific instructions provided to ChatGPT before the simulated conversation even began. Along the way, ChatGPT keeps a running short-term memory (called the "context window") of everything it and you have written, and when it "talks" to you, it is attempting to complete the transcript of a conversation as a text-completion task.

A diagram showing how GPT conversational language model prompting works.

Benj Edwards / Ars Technica

Additionally, ChatGPT is different from vanilla GPT-3 because it has also been trained on transcripts of conversations written by humans. "We trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides—the user and an AI assistant," wrote OpenAI in its initial ChatGPT release page. "We gave the trainers access to model-written suggestions to help them compose their responses."

ChatGPT has also been tuned more heavily than GPT-3 using a technique called "reinforcement learning from human feedback," or RLHF, where human raters ranked ChatGPT's responses in order of preference, then fed that information back into the model. Through RLHF, OpenAI was able to instill in the model the goal of refraining from answering many questions it cannot answer reliably. This has allowed the ChatGPT to produce coherent responses with fewer confabulations than the base model. But inaccuracies still slip through.

Why ChatGPT confabulates

Natively, there is nothing in a GPT model's raw data set that separates fact from fiction. That guidance comes from a) the prevalence of accurate content in the data set, b) recognition of factual information in the results by humans, or c) reinforcement learning guidance from humans that emphasizes certain factual responses.

The behavior of LLMs is still an active area of research. Even the researchers who created these GPT models are still discovering surprising properties of the technology that no one predicted when they were first developed. GPT's abilities to do many of the interesting things we are now seeing, such as language translation, programming, and playing chess, were a surprise to researchers at one point (for an early taste of that, check out 2019's GPT-2 research paper and search for the term "surprising").

So when we ask why ChatGPT confabulates, it's difficult to pinpoint an exact technical answer. And because there is a "black box" element of the neural network weights, it's very difficult (if not impossible) to predict their exact output given a complex prompt. Still, we know some basic things about how why confabulation happens.

Key to understanding ChatGPT's confabulation ability is understanding its role as a prediction machine. When ChatGPT confabulates, it is reaching for information or analysis that is not present in its data set and filling in the blanks with plausible-sounding words. ChatGPT is especially good at making things up because of the superhuman amount of data it has to work with, and its ability to glean word context so well helps it place erroneous information seamlessly into the surrounding text.

"I think the best way to think about confabulation is to think about the very nature of large language models: The only thing they know how to do is to pick the next best word based on statistical probability against their training set," said Simon Willison, a software developer who often writes on the topic.

In a 2021 paper, a trio of researchers from the University of Oxford and OpenAI identified two major types of falsehoods that LLMs like ChatGPT might produce. The first comes from inaccurate source material in its training data set, such as common misconceptions (e.g., "eating turkey makes you drowsy"). The second arises from making inferences about specific situations that are absent from its training material (data set); this falls under the aforementioned "hallucination" label.

Whether the GPT model makes a wild guess or not is based on a property that AI researchers call "temperature," which is often characterized as a "creativity" setting. If the creativity is set high, the model will guess wildly; if it's set low, it will spit out data deterministically based on its data set.

Recently, Microsoft employee Mikhail Parakhin, who works on Bing Chat, tweeted about Bing Chat's tendency to hallucinate and what causes it. "This is what I tried to explain previously: hallucinations = creativity," he wrote. "It tries to produce the highest probability continuation of the string using all the data at its disposal. Very often it is correct. Sometimes people have never produced continuations like this."

Parakhin said that those wild creative leaps are what make LLMs interesting. "You can clamp down on hallucinations, and it is super-boring," he wrote. "[It] answers 'I don't know' all the time or only reads what is there in the Search results (also sometimes incorrect). What is missing is the tone of voice: it shouldn't sound so confident in those situations."

Balancing creativity and accuracy is a challenge when it comes to fine-tuning language models like ChatGPT. On the one hand, the ability to come up with creative responses is what makes ChatGPT such a powerful tool for generating new ideas or unblocking writer's block. It also makes the models sound more human. On the other hand, accuracy to the source material is crucial when it comes to producing reliable information and avoiding confabulation. Finding the right balance between the two is an ongoing challenge for the development of language models, but it's one that is essential to produce a tool that is both useful and trustworthy.

There's also the issue of compression. During the training process, GPT-3 considered petabytes of information, but the resulting neural network is only a fraction of that in size. In a widely read New Yorker piece, author Ted Chiang called this a "blurry JPEG of the web." That means a large portion of the factual training data is lost, but GPT-3 makes up for it by learning relationships between concepts that it can later use to reformulate new permutations of these facts. Like a human with a flawed memory working from a hunch of how something works, it sometimes gets things wrong. And, of course, if it doesn't know the answer, it will give its best guess.

We cannot forget the role of the prompt in confabulations. In some ways, ChatGPT is a mirror: It gives you back what you feed it. If you feed it falsehoods, it will tend to agree with you and "think" along those lines. That's why it's important to start fresh with a new prompt when changing subjects or experiencing unwanted responses. And ChatGPT is probabilistic, which means it's partially random in nature. Even with the same prompt, what it outputs can change between sessions.

All this leads to one conclusion, one that OpenAI agrees with: ChatGPT as it is currently designed, is not a reliable source of factual information and cannot be trusted as such. "ChatGPT is great for some things, such as unblocking writer's block or coming up with creative ideas," said Dr. Margaret Mitchell, researcher and chief ethics scientist at AI company Hugging Face. "It was not built to be factual and thus will not be factual. It's as simple as that."

Can the fibbing be fixed?

Trusting an AI chatbot's generations blindly is a mistake, but that may change as the underlying technology improves. Since its release in November, ChatGPT has already been upgraded several times, and some upgrades included improvements in accuracy—and also an improved ability to refuse to answer questions it doesn't know the answers to.

So how does OpenAI plan to make ChatGPT more accurate? We reached out to OpenAI multiple times on this subject over the past few months and received no response. But we can pull out clues from documents OpenAI has released and news reports about the company's attempts to guide ChatGPT's alignment with human workers.

As previously mentioned, one of the reasons why ChatGPT has been so successful is because of extensive training using RLHF. As OpenAI explains, "To make our models safer, more helpful, and more aligned, we use an existing technique called reinforcement learning from human feedback (RLHF). On prompts submitted by our customers to the API, our labelers provide demonstrations of the desired model behavior and rank several outputs from our models. We then use this data to fine-tune GPT-3."

OpenAI's Sutskever believes that additional training through RLHF can fix the hallucination problem. "I'm quite hopeful that by simply improving this subsequent reinforcement learning from human feedback step, we can teach it to not hallucinate," Sutskever said in an interview with Forbes earlier this month.

He continued:

The way we do things today is that we hire people to teach our neural network to behave, to teach ChatGPT to behave. You just interact with it, and it sees from your reaction, it infers, oh, that's not what you wanted. You are not happy with its output. Therefore, the output was not good, and it should do something differently next time. I think there is a quite high chance that this approach will be able to address hallucinations completely.

Others disagree. Yann LeCun, chief AI scientist at Meta, believes hallucination issues will not be solved by the current generation of LLMs that use the GPT architecture. But there is a quickly emerging approach that may bring a great deal more accuracy to LLMs with the current architecture.

"One of the most actively researched approaches for increasing factuality in LLMs is retrieval augmentation—providing external documents to the model to use as sources and supporting context," said Goodside. With that technique, he explained, researchers hope to teach models to use external search engines like Google, "citing reliable sources in their answers as a human researcher might, and rely less on the unreliable factual knowledge learned during model training."

Bing Chat and Google Bard do this already by roping in searches from the web, and soon, a browser-enabled version of ChatGPT will as well. Additionally, ChatGPT plugins aim to supplement GPT-4's training data with information it retrieves from external sources, such as the web and purpose-built databases. This augmentation is similar to how a human with access to an encyclopedia will be more factually accurate than a human without one.

Also, it may be possible to train a model like GPT-4 to be aware of when it is making things up and adjust accordingly. "There are deeper things one can do so that ChatGPT and similar are more factual from the start," said Mitchell, "including more sophisticated data curation and the linking of the training data with 'trust' scores, using a method not unlike PageRank... It would also be possible to fine-tune the model to hedge when it is less confident in the response."

So while ChatGPT is currently in hot water over its confabulations, there may be a way out ahead, and for the sake of a world that is beginning to rely on these tools as essential assistants (for better or worse), an improvement in factual reliability cannot come soon enough.

Why ChatGPT and Bing Chat are so good at making things up

Google CEO says conversational AI will come to Search

Thu, 06 Apr 2023 15:56:57 +0000

Google’s CEO, Sundar Pichai, has confirmed in an interview with The Wall Street Journal that Google Search will start to receive conversational artificial intelligence elements in upcoming updates to compete with Bing Chat, which Microsoft has been offering to customers for a while now. Microsoft has even started rolling Bing Chat out to other services including SwiftKey Beta.

In the interview with WSJ, Pichai said that AI will supercharge Google Search’s ability to answer users’ requests and said that rival chatbots like Bing Chat and ChatGPT do not pose a threat to Google Search. He went on to say:

“The opportunity space, if anything, is bigger than before. Will people be able to ask questions to Google and engage with LLMs in the context of search? Absolutely.”

In another recent interview, Pichai said that Google Bard, its generative AI project, would switch over to a new large language model (LLM) called PaLM. The Google CEO said that PaLM is a more capable model, and according to Bard, has been shown to be better at some tasks such as code generation and translation while LaMDA is better at question answering and creative writing.

Source: WSJ

Source

Italy’s ChatGPT ban spreads to France, Germany and Ireland

Thu, 06 Apr 2023 15:46:51 +0000

Italy’s ChatGPT ban has caused numerous European countries to focus on banning the AI platform.

Germany, France and more are looking to impose a ChatGPT ban over numerous privacy concerns. Following in the footsteps of Italy’s Data Protection Authority, a number of European companies have it out for generative AI.

Via tech news outlet Reuters, it was reported that multiple countries are looking to enforce bans on the OpenAI platform. In an effort to protect individual privacy, ChatGPT may be banned across most of Europe.

The report explains that German data protection authorities are already looking to join Italy in enforcing a ban. Speaking to German newspaper Handelsblatt, the commissioner for Germany’s privacy watchdog revealed that there are discussions to ban the service in the country.

Ireland's Data Protection Commissioner also confirmed that they are in discussions with Italy regarding the ChatGPT ban.

“We are following up with the Italian regulator," a spokesperson for the initiative said. "We will coordinate with all EU data protection authorities in relation to this matter."

Italy’s ban of OpenAI’s generative chatbot targets the company’s recent data breach as well as the vast amounts of data it collects to train its AI program. The data protection watchdog has claimed that it will lift the ban if OpenAI can protect Italians’ data.

However, the move has been met with a large degree of criticism. Even government members in Italy believe that the ban is too much and is built on little evidence.

Furthermore, only OpenAI’s standalone version of the software has been banned in the country. Microsoft’s Bing AI, which is powered by ChatGPT, has not been banned despite being the same service in all intents and purposes.

There are huge discussions that need to be had surrounding the future of AI services. With AI still vastly unregulated, the technology does pose dangers towards industry, the economy and even mental health.

Source

Move Aside 7950X3D, AMD Ryzen 7800X3D Is New Gaming CPU King

Thu, 06 Apr 2023 04:18:54 +0000

7950X3D vs 7800X3D vs 13900K: The reviews for the AMD Ryzen 7 7800X3D are out. Almost all of them show it beating other offerings from AMD and Intel while costing far cheaper.

When AMD announced Ryzen 7000 series, everyone, especially gamers, were waiting for one thing. Whether AMD is going to release a 3D V-Cache based CPU in it too. Many wanted AMD to release a Ryzen 7000 series version of massively popular AMD Ryzen 7 5800X3D.

To everyone’s surprise, AMD announced not one, but three processors with 3D V-Cache. AMD Ryzen 9 7950X3D, 7900X3D and Ryzen 7 7800X3D. This took the amount of X3D chips that AMD offered to 4 in total so far.

What 3D V-Cache does is it installs an additional 64 MB of cache on top of a pre-existing 32 MB cache on a single CPU chip. This gives the extra cache based CPUs a tremendous speed boost in workloads which benefit from it. Which is mostly limited to gaming.

When AMD released Ryzen 9 7950X3D and Ryzen 9 7900X3D, many wondered about two things. First, how will AMD Ryzen 7 7800X3D perform. Second, how will AMD install the extra cache on the top two CPUs.

The reason for the above question is simple. While AMD Ryzen 7 7800X3D comes with a single 8-core chiplet. AMD Ryzen 9 7950X3D and 7900X3D come with two 8-core and 6-core chiplet respectively. So What AMD did was to keep one chiplet without extra cache and boost its clock speed. The other chiplet, however, came with extra cache like mentioned above. The reason is simple, games don’t benefit much for above 8 cores.

So when Ryzen 7 7800X3D was announced. Everyone got excited. Looks like the reviews are exciting too.

Tom’s Hardware

AMD Ryzen 7800X3D FPS Review Benchmark Average FPS. Credit: Tom’s Hardware.

Tom’s Hardware did a review for the AMD Ryzen 7 7800X3D.

First of all. They mentioned how the chipset driver AMD had released for Ryzen 9 7950X3D and 7900X3D created issues with Ryzen 7 7800X3D. This is because it was wrongly parking CPU cores in the 7800X3D during gaming (page 3 of the review), something which was meant for parking non-3D Cache chiplet cores in Ryzen 9 7950X3D and 7900X3D processors. The only way to fix this was to re-install the OS completely from the start.

Talking about the review. AMD Ryzen 7 7800X3D performs mediocre or poorly in productivity software due to it being a lower clocked CPU. However, what it excels is in gaming. In gaming benchmarks, AMD Ryzen 7 7800X3D equals or beats 7950X3D by a small margin at 1080p. Not only that, it’s 11% faster than the Intel’s flagship offering, the Intel Core i9-13900K at 1080p.

What is interesting is that the review finds AMD Ryzen 7 7800X3D to be significantly slower than Ryzen 9 7950X3D and Intel Core i9-13900K in 99th percentile FPS at 1080p, which is important for smoothness of gameplay. The reviewer blame a single game for the anomaly. However, that is fixed in the 1440p where Ryzen 7 7800X3D is back at top in 99th percentile FPS. The average FPS is a non-issue though, Ryzen 7 7800X3D is absolutely on top.

AMD Ryzen 7800X3D Power Usage Watts. Credit: Tom’s Hardware.

In terms of power usage (page 2), AMD Ryzen 7 7800X3D is extremely efficient. At max, it uses 82W at stock. Whereas Ryzen 9 7950X3D uses 143W and Intel Core i9-13900K uses a huge 280W for the same test.

igor’sLAB

AMD Ryzen 7800X3D FPS Review Benchmark Average FPS. Credit: Igor’sLAB.

igor’sLAB too did a review of AMD Ryzen 7 7800X3D.

In performance in 1080p (page 4), AMD Ryzen 9 7950X3D beats AMD Ryzen 7 7800X3D in average FPS tests by a very small and insignificant margin. Which they blame on slightly lower clock rates. However, that changes in 1440p tests where 7800X3D is back on top.

Here too in 1% low FPS, the Ryzen 9 7800X3D lags behind others a little at 1080p and 4K, but improves on 1440p.

Either way, at 1080p, AMD Ryzen 9 7800X3D is 9% faster than Intel Core i9-13900K, 17% faster than AMD Ryzen 5800X3D and 26% faster than the $50 cheaper Ryzen 7 7700X.

AMD Ryzen 7800X3D Gaming 1080p Power Usage In Watts. Credit: Igor’sLAB.

In terms of power usage (page 11), here too, AMD Ryzen 9 7800X3D is extremely efficient. At max using 88W in complex applications whereas Ryzen 9 7950X3D uses 144W and Intel Core i9-13900K uses a huge 280W for the same test. They however checked power usage in gaming too. In gaming, AMD Ryzen 9 7800X3D averaged at 40W, while Intel Core i9-13900K used 75W.

TechPowerUp

AMD Ryzen 7800X3D Average FPS Review Benchmark. Credit: TechPowerUP.

TechPowerUp was another one with the review for the AMD Ryzen 7 7800X3D.

They too mentioned the problem they had with AMD chipset drivers and core parking (page 28) and they too had to reinstall the OS to get rid of the issue, as recommended by the AMD’s guide.

In terms of gaming performance. Here too, the AMD Ryzen 9 7800X3D shines a lot. At 1080p, it’s almost 6% faster than both AMD Ryzen 9 7950X3D and Intel Core i9-13900K. At 1440p and 4K too it’s ahead than others.

In minimum FPS tests (page 22), AMD Ryzen 9 7800X3D leads ahead at 1080p, while being only slightly slower (1 FPS) than the other two at 4K.

In power efficiency (page 23), the power usage was similar as other reviewers in applications. However, while in gaming where AMD Ryzen 9 7800X3D used just 49W, Intel Core i9-13900K used 143W, almost three times more.

What TechPowerUp did which many other reviews didn’t. They tested the temps of AMD Ryzen 9 7800X3D. At stock, it didn’t go above 82C in complex workload and the gaming temps were significantly lesser than that.

TechSpot

AMD Ryzen 7800X3D Average FPS Review Benchmark. Credit: TechSpot.

TechSpot posted their review for AMD Ryzen 9 7800X3D only slightly later than others.

In performance at 1080p, it scores the same as AMD Ryzen 9 7950X3D. It’s also 3% faster than Intel Core i9-13900K.

AMD Ryzen 7800X3D Power Usage Watts. Credit: TechSpot.

In total PC power usage, the 7800X3D PC uses 207W. Whereas, AMD Ryzen 9 7950X3D uses 279W and Intel Core i9-13900K uses a huge 493W for the same. This information is important for those who buy a 1000W or more PSU for their PC.

Conclusion

There are many other reviewers one can check out. Like the detailed written ones from AnandTech. There is one that KitGuru has gotten too. Even famous YouTube reviewers like Gamers Nexus and Hardware Unboxed have posted their reviews on it.

Whatever review one checks, overall they all say the same thing. AMD Ryzen 7800X3D is as fast as or faster than the AMD Ryzen 9 7950X3D. It also beats Intel Core i9-13900K by some small margin. However, what’s the biggest difference is that it’s extremely power efficient than others.

It’s important to mention that a $449 AMD Ryzen 7800X3D is equaling a $699 AMD Ryzen 9 7950X3D and beats a $589 Intel Core i9-13900K.

Talking about the costs. The only issue until recently was the platform as a whole. The AM5 motherboards required for it are expensive. The DDR5 RAM which it requires were expensive too.

However, both the things can be largely sorted now. Higher quality versions of recently released A620 chipset motherboards can be paired with this CPU, though after some checking as some A620 ones only allow 65W CPU, whereas AMD Ryzen 7800X3D is rated for 120W. So it’s important to check the specs. The DDR5 RAM prices too have gotten cheaper considerably recently.

Some people complain about productivity speeds in this CPU. Which is understood if someone uses these things professionally. For those type of users, AMD Ryzen 7950X3D is suggested. However, common gamers are going to benefit massively from the Ryzen 7800X3D. Not to forget, speed difference in common tasks are hardly noticeable.

In all, AMD Ryzen 7800X3D is one of the best and greatest CPUs ever created in the world of computer gaming. Let’s hope Intel too has an answer for it. Maybe they too can consider adding extra cache in their processors.

Move Aside 7950X3D, AMD Ryzen 7800X3D Is New Gaming CPU King

Asus AGESA BIOS for AMD Ryzen 7000 reduces massively long Windows boot times, updates TPM

Wed, 05 Apr 2023 20:18:06 +0000

While AMD's Ryzen 7000 series processors are really efficient processors, especially the newer X3D chips, many users are perplexed by the slow boot times on their system as generally, a new generation of product is typically always faster than last gen stuff, or at least, on par. That is often not the case with AMD's Zen 4 chips, at least when it comes to booting.

But things are starting to get better it seems. Asus recently released a firmware update for its ROG X670E HERO motherboard and Twitter user HXL notes that the new AMD Generic Encapsulated Software Architecture (AGESA) 1.0.0.6 update reduced their boot times from 56 seconds down to 30 seconds which is nearly half. This was measured on the new Ryzen 7950X3D processor running dual 32GB DDR5-5600 kit.

The changelog for the latest firmware (version 1004) notes:

Version 1004

Update AGESA version to ComboAM5PI 1.0.0.6
TPM 2.0 security update

As you may have noticed above, it looks like Asus has made some updates related to TPM 2.0 as well though the changes made have not been explicitly stated. Alongside these improvements, the previous 1003 beta version also added support for high-density DDR5 modules which may explain the improved boot times too.

Version 1003 Beta Version

Update AGESA version to ComboAM5PI 1.0.0.6
Supports high density DDR5 module
TPM 2.0 security update

In case you are wondering how the boot times from the Asus ROG X670E Hero compare to those from other vendors, the new lowered boot time certainly brings it in the middle of the stack now. Some ASRock and Gigabyte motherboards seem to be the fastest in this regard (via Hardware Unboxed YouTube channel):

Alongside slow boot times, AMD Ryzen 7000 processors also have a really slow initial booting speed especially when the memory DIMMs are fully populated.

For example, ASRock notes (image above) that a quad 32GB kit can take up to around 400 seconds (or close to 7 minutes) for the initial boot after the CMOS has been reset. This time is taken up for the purpose of DDR5 "memory training" on the motherboards.

Asus AGESA BIOS for AMD Ryzen 7000 reduces massively long Windows boot times, updates TPM

No Man's Sky gets new ship class, Xbox visual upgrades, corrupted planets and more

Wed, 05 Apr 2023 20:13:00 +0000

It was only February when Hello Games shipped its massive Fractal update to No Man's Sky, but already, another major content drop is here. Dubbed the Interceptor Update, it adds a twist to the Sentinel drones that have been policing the universe, a fresh type of planet to explore, and there's good news for Xbox players.

Players are now able to find crashed Sentinel Interceptor starships in the wild to repair and add to their fleet. These come with a unique vestigial cockpit design as well as a new planetary flight model. The ships also tout unique technologies as upgrades to improve their behavior.

At the same time, some heavily fortified Sentinel planets have become corrupted worlds. These dangerous environments have upgraded robotic guardians to defeat, such as "heavily-armored semi-arachnid machines", drones with flamethrowers and grenade launchers, and more. Those who are brave enough to venture in can get corrupted resources, find abandoned encampments to explore the story, and possibly even run across an Interceptor to claim.

Xbox console players will also find the game looking better than before. Hello Games is utilizing dynamic resolution scaling and AMD's FSR 2.0 upscaling tech to improve visual fidelity. This will be most noticeable during spaceflight, per the studio. Read the full patch notes at the bottom of the announcement page here.

The No Man's Sky Interceptor update is now available across PC, Xbox One, Xbox Series X|S, Game Pass, PlayStation 4, PlayStation 5, and Nintendo Switch systems.

No Man's Sky gets new ship class, Xbox visual upgrades, corrupted planets and more

Google claims its new TPU-based supercomputer can beat NVIDIA's AI chips

Wed, 05 Apr 2023 20:11:04 +0000

Google wants to prove that it can make a supercomputer that can handle the growing number of generative AI applications. Today, Google researchers published a paper (via CNBC) that says its TPU v4 supercomputer is "1.2x–1.7x faster and uses 1.3x–1.9x less power" compared to systems that used NVIDIA's A100 chip.

Google is trying to compete with NVIDIA in this market, where it is playing catch up. NVIDIA currently dominates in this sector with over 90 percent of AI-based development using its chips. Google wants companies to use supercomputers that use its Tensor Processing Units (TPU) instead. In fact, Google said that Midjourney, the popular AI-generated image maker, uses TPU-based chips.

While the paper compares the TPU v4 to the A100, Google said it did not compare the supercomputer to ones that use NVIDIA's H100 chips. These are the same processors that Microsoft is using for its own AI applications, including Bing Chat and Microsoft 365 Copilot.

It's clear that Google feels that AI and the supercomputers that power those applications are the future of computing and technology. However, even with this new TPU v4 setup, it will likely struggle to compete with NVIDIA's lead in this matter.

Google claims its new TPU-based supercomputer can beat NVIDIA's AI chips

Should AI build your next PC for you?

Wed, 05 Apr 2023 20:10:23 +0000

Using Newegg's recently launched AI-powered PC Builder should make the PC building process a lot easier. That's at least Newegg's promise. Launched as a Beta, PC Builder with AI takes the user's instructions to find suitable components for building a PC.

All that is required is to feed it with instructions, say "PC for playing games, price $1400-$1700, AMD processor", and wait for the AI to come up with components that match the instructions. If it works, it would make the PC building process a lot easier for users who may not want to pick components manually after careful research.

The idea makes a lot of sense, but everything depends on the execution. Newegg uses ChatGPT to power the AI-based suggestions. To get started, head over to the main Newegg website and activate the PC Builder link at the top.

Type instructions into the search field and hit the "Build with AI" button afterwards. The AI analyzes the input and returns sample PC builds and components. In an ideal world, the PCs would match the user's instructions exactly.

In reality, Newegg's AI may already stumble here. While the PCs look good on first glance, users may notice that at least some of them may be over budget. The PC Builder did take other instructions into account most of the time, e.g., that AMD systems were suggested if the instructions requested that. Sometimes, though, PC builder would interpret the instructions differently. Some of these issues may be corrected by adding details to the instructions. Simply stating AMD may, for instance, return a PC with an Intel processor and an AMD graphics card. Rephrasing this to AMD CPU and graphics card, did return the desired components.

Some of the AI's suggestions were puzzling. Inexperienced users may not notice these issues. In one instance, PC Builder suggested DDR4 RAM for a DDR5 motherboard, in another, a CPU that was not current gen, even though instructions requested the system to have a current generation PC.

The PC builds that the AI suggested should all work technically, but the configuration was in several cases not optimized or ideal. The suboptimal components were not picked because of budget constraints either in most cases.

Lastly, users need to verify if the selected components fit inside the case; this is especially important for video cards, which may require lots of space, but also other components.

Closing Words

Newegg's PC Builder is labeled Beta, and there is certainly a reason for that. Users interested in building a PC may find the new AI-powered feature useful, but it is important to verify the suggestions and not blindly trust them.

With that said, PC Builder powered by AI could one day become a useful tool for PC builders. For now, it is not yet ready for the task.

Now You. do you build your own PCs?

Should AI build your next PC for you?

More layoffs at Amazon: Over 100 employees let go

Wed, 05 Apr 2023 20:09:05 +0000

Amazon has laid off thousands of its employees in the past year and continues to do so. Recent reports show that the company has cut over 100 employees working in the gaming divisions.

Amazon Games VP Christoph Hartmann wrote a memo to staff on Tuesday, giving more information about the planned layoffs. Hartmann said the decision was made to restructure and streamline the company's operations.

"After evaluating our current projects against our long-term goals, the Games leadership team made the difficult decision to eliminate just over 100 roles across Prime Gaming, Game Growth, and in our San Diego studio while also reassigning some employees to other projects that match our strategic focus," Hartmann wrote.

Amazon Prime Gaming

Part of 9,000 layoffs announced earlier

The layoffs are a part of 9,000 job cuts announced a couple of months ago. Amazon started cutting the workforce from its gaming divisions. It also parted ways with 400 employees working for Twitch, and now more than 100 are left in limbo in Amazon's own gaming division.

Amazon's gaming division is working on new projects since 2013, but the team has failed to meet expectations. They developed and published New World, which was definitely a hit but got criticized at the launch due to bugs and other issues. The company has also published Lost Ark, a game that Smilegate and Tailpod Studio developed.

"Going forward, we will continue to invest in our internal development efforts, and our teams will continue to grow as our projects progress," Hartmann added.

Employees affected by this decision were given the information in live meetings scheduled for Tuesday morning with human resources. The company also added that it would give additional benefits to those who lost their jobs, including severance pay, health insurance benefits, outplacement services, and paid time to conduct their job search.

The tech giant also lost two key employees, Mike Frazzini, who helped launch Amazon Games, and Jhon Smedley, the head of Amazon Games' San Diego studio. Frazzini left the company last March, and Smedley stepped down in January.

Amazon is one of the biggest technology companies in the world. On the other hand, it is also one of the companies that laid off thousands of employees quickly. It says that these changes are mandatory to streamline its operations and have a stable future despite global economic difficulties.

More layoffs at Amazon: Over 100 employees let go

10x EV Range Boost With Revolutionary Lithium-Ion Battery Technology

Wed, 05 Apr 2023 18:04:23 +0000

POSTECH-Sogang University joint research team develops layering-charged, polymer-based stable high-capacity anode material.

The electric vehicle market has been experiencing explosive growth, with global sales surpassing $1 trillion (approximately 1,283 trillion Korean Won/KRW) in 2022 and domestic sales exceeding 108,000 units. Inevitably, demand is growing for high-capacity batteries that can extend EV driving range. Recently, a joint team of researchers from POSTECH and Sogang University developed a functional polymeric binder for stable, high-capacity anode material that could increase the current EV range at least 10-fold.

A research team led by POSTECH professors Soojin Park (Department of Chemistry) and Youn Soo Kim (Department of Materials Science and Engineering) and Professor Jaegeon Ryu (Department of Chemical and Biomolecular Engineering) of Sogang University developed charged polymeric binder for a high-capacity anode material that is both stable and reliable, offering a capacity that is 10 times or higher than that of conventional graphite anodes. This breakthrough was achieved by replacing graphite with Si anode combined with layering-charged polymers while maintaining stability and reliability. The research results were published as the Front Cover Article in Advanced Functional Materials.

Credit: POSTECH

High-capacity anode materials such as silicon are essential for creating high-energy density lithium-ion batteries; they can offer at least 10 times the capacity of graphite or other anode materials now available. The challenge here is that the volume expansion of high-capacity anode materials during the reaction with lithium poses a threat to battery performance and stability. To mitigate this issue, researchers have been investigating polymer binders that can effectively control the volumetric expansion.

However, research to date has focused solely on chemical crosslinking and hydrogen bonding. Chemical crosslinking involves covalent bonding between binder molecules, making them solid but has a fatal flaw: once broken, the bonds cannot be restored. On the other hand, hydrogen bonding is a reversible secondary bonding between molecules based on electronegativity differences, but its strength (10-65 kJ/mol) is relatively weak.

The new polymer developed by the research team not only utilizes hydrogen bonding but also takes advantage of Coulombic forces (attraction between positive and negative charges). These forces have a strength of 250 kJ/mol, much higher than that for hydrogen bonding, yet they are reversible, making it easy to control volumetric expansion. The surface of high-capacity anode materials is mostly negatively charged, and the layering-charged polymers are arrayed alternately with positive and negative charges to effectively bind with the anode. Furthermore, the team introduced polyethylene glycol to regulate the physical properties and facilitate Li-ion diffusion, resulting in the thick high-capacity electrode and maximum energy density found in Li-ion batteries.

Professor Soojin Park explained, “The research holds the potential to significantly increase the energy density of lithium-ion batteries through the incorporation of high-capacity anode materials, thereby extending the driving range of electric vehicles. Silicon-based anode materials could potentially increase driving range at least tenfold.”

Source

Italy's ChatGPT ban fails to deter users due to VPNs

Wed, 05 Apr 2023 17:43:42 +0000

VPN applications have gained a vast amount of new users after Italy's ChatGPT ban. People are using the apps to access the chatbot.

Recently, the Italian government banned ChatGPT due to privacy reasons. However, this didn't stop people from reaching OpenAI's services. PureVPN has noticed an odd increase in traffic coming from Italy on their website after the ban went into effect on April 1. According to a recent blog post from the company, "Italians have been turning to VPN services following the decision of the country's data protection authority to ban ChatGPT over privacy concerns."

PureVPN

ChatGPT is one of the most popular topics on the internet. People from all around the globe use it to get their work done easier. However, the Italian government prevented the chatbot from being used in the country. Authorities further alleged that OpenAI failed to verify its users' ages and enforce prohibitions barring anyone under the age of 13 from using ChatGPT.

"ChatGPT has garnered more than 100 million users since its launch two months ago. The advanced chatbot is just as popular in Italy as in other countries because of its ability to have human-like conversations. However, with Italians unable to access ChatGPT, many of them are turning to VPNs to circumvent the block," says PureVPN.

VPNs can help users mask their real IP addresses and use a different one from their selected country. Italian people use VPNs and an IP address of another country to access ChatGPT.

PureVPN

What happened before?

The Italian government issued a country-wide ban on ChatGPT because of the "unlawful collection of personal data." Italy accused OpenAI of privacy issues and ordered it to stop collecting data from Italian people. The software was banned after the Garante,

Italy's data protection watchdog, charged OpenAI of inappropriately gathering and retaining Italians' personal data and, as a result, violating GDPR rules.

Italy is the first Western country to ban the chatbot. Countries like Russia, China, North Korea, and Iran have already banned the chatbot, but Italy is the first in the West. OpenAI rejected the allegations and said it is protecting people's privacy and minimizing the collection of personal data.

"We also believe that AI regulation is necessary - so we look forward to working closely with the Garante and educating them on how our systems are built and used." a spokesperson of OpenAI told the BBC.

Other European countries may also issue similar bans in the coming weeks. Garante was one of the first organizations to report a potential breach of EU privacy regulations of TikTok. TikTok is now banned by the European Commission.

Source

Apple readies opening of its first retail store in India

Wed, 05 Apr 2023 13:43:15 +0000

Apple is gearing up to open its first retail store in India to the public later this month, roughly two decades after the company began selling products and services in the South Asian nation that has grown to become the second-largest internet market.

The iPhone-maker on Wednesday shared the barricade of its first retail store in India, called Apple BKC, situated at Jio World Drive Mall in Mumbai.

“Inspired by the iconic Kaali Peeli taxi art unique to Mumbai, the Apple BKC creative includes colourful interpretations of the decals combined with many Apple products and services that will be available for our customers to discover,” the company said in a statement.

Apple also plans to launch its second India store somewhere in New Delhi in the coming quarters.

Apple has been aggressively hiring employees for its stores in India in recent months, according to recruitment posts. The company, which launched its online store in India in 2020, had initially planned to debut the first retail outlet in 2021 but delayed those plans after COVID struck.

India, the world’s second-largest smartphone market, currently contributes little to Apple’s topline, but the iPhone maker has said in the past that it’s bullish about its prospects in the country.

Apple’s contract manufacturing partners, Foxconn and Wistron, have ramped up local assembling for iPhone and other Apple gadgets in recent quarters. In a report last year, JP Morgan analysts estimated that Apple would expand its manufacturing capacity in India to produce 25% of all iPhones by 2025.

Source

Microsoft rolling out Bing Chat v98 with less refusals, and reduced disengagement

Wed, 05 Apr 2023 13:26:03 +0000

Microsoft's Bing Chat team isn't taking time off this week as Easter weekend approaches. Early this morning, Mikhail Parakhin, Microsoft's head of Advertising and Web Services, posted on his Twitter account that v98 of Bing Chat should start rolling out today. The company released v96 back on February 28.

Starting to ship Prompt v98 today: it is a two-stage process, by tomorrow you should see big reduction in the number of cases when Bing Chat refuses to create something (write code, for example). Then the second stage will be deployed, reducing disengagements.

— Mikhail Parakhin (@MParakhin) April 5, 2023

In his post, Parakhin stated the release of Bing Chat v98 will be in two stages. The first stage is designed to give Bing Chat users large reductions "in the number of cases when Bing Chat refuses to create something" according to Parakhin. The second stage will then roll out which should cut back on the chatbot's disengagements.

Parakhin also responded to some messages from Twitter users. One complained that the recent release of the Bing Image Creator was still "very restrictive" in terms of making AI-generated art. Parakhin replied, "Also rapidly improving - it is much better than when we first released it."

Another user stated that Bing Chat in its Creative mode was not working correctly, offering incorrect answers to questions. Parakhin stated, "

Sorry about that. We are trying to have faster responses: have two pathways in the model, run the fast one, check if the answer is good, if not - run the slower pathway. In this case it decided incorrectly that the additional wait is unwarranted 😞 We'll fix it.

He later expanded on that answer, stating:

Our push is to try to bring the best user experience to everyone. It's not like we are holding back - the case above wasn't to save capacity, it was an attempt to speed up the Creative mode without regressing on quality. We are not perfect. Yet 🙂

The Bing Chat team was busy last week with some new improvements and features, including expanding its daily and per-session chat turn limits, and adding a way to search for images and video inside the chat dialog.

Source

Microsoft is 48 years old today. Here's a quick look back at its creation

Wed, 05 Apr 2023 05:05:42 +0000

In 1975, two men formed a new tech company in a rather unlikely location. One founder was just 19 years old and had dropped out of Harvard University. The other founder actually convinced his friend to drop out of Harvard, just as he left Washington State University. The two men moved to Albuquerque, New Mexico, and started their new business.

Those two men were Bill Gates and the late Paul Allen. On April 4, 1975, they officially formed Micro-Soft. Neither one likely had any idea that their company would become one of the biggest and most influential companies ever created.

Today is, of course, the 48th birthday of Microsoft. The company is not celebrating the event with a lot of fanfare, but it is noting the day on its official Twitter account with a retro-style logo.

Here are a few more interesting tidbits about the founding of Microsoft.

The proto-Microsoft: Traf-O-Data

Before the launch of Microsoft, Gates and Allen, along with another person, Paul Gilbert, founded a company called Traf-O-Data in Seattle, Washington. Gates and Allen developed software for the company, and Gilbert built a computer for it. In a 1993 interview, Gates talked about the idea behind the company:

Traf-O-Data was taking road volume data and converting it into reports where you have actually just a 16-channel paper tape on the side of the road. The pressure sensitive hose that you drive over has a counter in there clicking out a count every five, ten, or fifteen minutes. Those have to be processed for the State Road Departments, to give out money for repairs, and decide how to do traffic lights, etc. Anyway, it was data that needed to be processed. So, we got involved in that.

Gates said in the same interview that they "made a little bit of money and had some fun with it." However, that business came to a fairly quick end after the state of Washington started processing this traffic data for free for the state's cities.

Why was Microsoft founded?

Both Gates and Allen learned about the world's first microcomputer, the Altair 8800, in late 1974 by reading an article about it in Popular Electronics magazine. It was made by a company called Micro Instrumentation and Telemetry Systems (MITS) in Albuquerque, New Mexico. Gates and Allen thought they could create a version of BASIC that would work on the Altair 8800. They moved to Albuquerque in 1975 to officially form Micro-Soft.

Where did the name come from?

The name Microsoft was Allen's idea, as he thought it was a good mashup of "microcomputer" and "software".

Where was the company founded?

Gates and Allen formed Microsoft in a room at the Sundowner Hotel, off the legendary Route 66. The Sundowner Hotel is still there, although it was converted to low-cost apartments in 2014.

Later in 1975, Gates and Allen moved the business to a proper office building in Albuquerque. In 2006, a plaque was put in place to commemorate that location, although it's been ripped off from its stone base at least once.

Why did Microsoft leave Albuquerque?

In short, Albuquerque was not the best location for recruiting people to join a growing tech company, so Gates and Allen decided to go back close to their hometown in Redmond, Washington, in 1979.

Happy 48th birthday, Microsoft! You have certainly come a very long way from Traf-O-Data and that room at the Sundowner Hotel. Here's to many more celebrations, including your milestone 50th anniversary in two years' time.

Microsoft is 48 years old today. Here's a quick look back at its creation

Graphics Card Maker Nvidia Was Founded 30 Years Ago

Tue, 04 Apr 2023 20:46:42 +0000

Nvidia as a company was invented 30 years ago on 5th April 1993. From starting as a small company made by 3 engineers, it has become a pioneer in GPU technology.

30 years ago, on 5th of April in the year 1993, three engineers were sitting inside a diner. During discussion, they realized that the computing abilities back then were not enough and graphics acceleration based computing was the way forward. When used in video games, it could solve many computing problems with high amount of sales expected. This is when Nvidia was founded.

The three people who founded Nvidia were no small timers. It included the now prominent face of Nvidia, it’s current CEO Jensen Huang, who was one of the directors at a semiconductor company LSI Logic and also worked as a microprocessor designer at AMD (yes, Jensen used to work for AMD when it was a CPU company).

The other two were Chris Malachowsky, who was an engineer at Sun Microsystems and Curtis Priem who was a graphics chip designer at IBM and Sun Microsystems. Together, they founded Nvidia with just $40,000 in their bank. With more funding coming only later.

Initially, Nvidia had no such name to it. They used to only use NV as their name, with NV being an abbreviation for Next Version. Then, after a lot of searching, an idea came from the Latin word invidia, which means envy, that the company’s name as we know it was selected.

Legacy

The Diamond’s Edge 3D 3240 Graphics Card with the NV1 chip. Credit: TechPowerUP.

A year later, in partnership with SGS-Thomson Microelectronics, Nvidia made it’s very first commercially available product. The Nvidia NV1. It was a $299 PCI based card on a 500nm process with 1 million transistors, with just 75 MHz GPU speed and 2 MB of memory. With a dozen of rival graphics companies around, the card wasn’t successful.

It was only in 1998 that Nvidia found some footing with their release of NVIDIA Riva TNT card. It was based on 350nm process, had 7 million transistors and came with a 90 MHz GPU clock and 16 MB memory.

Nvidia GeForce 256’s chip-shot.

However, the biggest product by Nvidia back then came a year later. Nvidia released GeForce 256, which it marketed as “world’s first GPU”. Made by TSMC, this was one of the graphics cards which ensured that Nvidia goes onto to become one of the biggest names in the video game hardware industry. Based, on a 220nm process, GeForce 256 came with 17 million transistors, 120 MHz GPU clock and two different types of memory, SDR and DDR. While SDR was slower and made by Samsung, a faster DDR version was made by HK Hynix (Hyundai Electronics back then). However, both came with 32 MB memory.

Later, Nvidia also got a contract to develop the graphics chip for the very first XBOX console. For which Nvidia got a huge $200 million in advance. Years later, it also got a contract to develop a graphics chip for PlayStation 3 console.

Meanwhile, Nvidia managed to acquire many companies, including 3dfx, which was one of the biggest names in the graphics card industry back then. Besides, it’s another rival, ATI, was sold to AMD in 2006.

Future Forward

Nvidia GeForce RTX 4090 Graphics Card. Credit: Nvidia.

Throughout the years, Nvidia has released many graphics cards. Too many of them to even list. The latest one being Nvidia GeForce RTX 4090 which has ensured that Nvidia remains the leader in the graphics card market out there.

Not to forget, few years ago Nvidia did make something standout. It introduced Ray Tracing hardware in its RTX graphics cards, starting with RTX 20 series, which was a new introduction in the world of graphics and gaming. It completely changed the lighting and reflections looked and worked in video games. Though it’s questionable whether the reduction in performance is worth it, thanks to heavy resources required to run it.

However, Nvidia of the present day isn’t concentrating on graphics cards alone. It makes portable gaming consoles in Nvidia SHIELD. Not to forget, it makes mobile GPU chips. Going forward, Nvidia wants to also make further improvements in AI chips, supercomputer modules, data-center among other things.

These days, Nvidia has almost become a monopoly in graphics card market, capturing a huge market share. With their only rivals, AMD and now latest entrant Intel, being far behind in market share than what they ideally should be, This has meant that Nvidia has been putting their recent generation of graphics cards at unbelievable and sometimes unacceptably high prices.

Whatever it maybe. Nvidia remains and might continue to be one of the biggest names in the gaming hardware industry. We wish them their 30th birthday and hope that they continue making great products, though price them affordably better too.

Graphics Card Maker Nvidia Was Founded 30 Years Ago

Bill Gates speaks out against artificial intelligence pause saying it won't solve challenges

Tue, 04 Apr 2023 18:21:24 +0000

Bill Gates has said a pause on the development of artificial intelligence won’t solve the challenges the technology presents, Reuters has reported. His comments come after 1,100 signatories published a letter calling for all generative AI labs to halt the training of AI systems for half a year for guardrails to be put in place.

On the matter of a pause, the Microsoft co-founder said:

“I don’t think asking one particular group to pause solves the challenges. Clearly there’s huge benefits to these things… what we need to do is identify the tricky areas. I don’t really understand who they’re saying could stop, and would every country in the world agree to stop, and why to stop. But there are a lot of different opinions in this area.”

When it was getting on for the end of March, Gates posted a piece on his blog explaining that artificial intelligence was as revolutionary as mobile phones and the internet. It’s true we’ve had AI for a while through services like Alexa and Google Assistant but with ChatGPT and Google Bard, we’ve seen a substantial improvement in the technology.

It’s not clear yet whether the letter calling for a pause will be heeded or ignored, but over the weekend, Italian deputy PM Matteo Salvini said the block on ChatGPT introduced by the country’s Data Protection Authority was excessive. This could suggest that politicians aren’t on board with limiting access or stifling these services.

Source: Reuters

Bill Gates speaks out against artificial intelligence pause saying it won't solve challenges

Google is rolling back the unannounced file limit change on Drive, promises to do better

Tue, 04 Apr 2023 18:20:18 +0000

Some Google customers have run into hard file limits on Google Drive for the past two months. User accounts that exceeded the hard limit of 5 million files could not add more files to cloud storage solution, even if enough storage space was available.

Google Drive's error message, "Error 403: This account has exceeded the creation limit of 5 million items. To create more items, move items to the trash and delete them forever", suggested to users that they should remove files to get under the 5 million files limit, if they wanted to add files to Google Drive.

Google did not document the change, and some Google Drive customers could not use the service as a consequence. Google never listed the file limit in the available plans, and there has been no support document available either that revealed the hard file cap.

Google did respond to an Arstechnica article two months later. In the statement, Google confirmed the existence of the file limit and stated that it was "a safeguard to prevent misuse of our system in a way that might impact the stability and safety of the system".

Today, Google announced that it made the decision to revert the change. The official Google Drive Twitter account published the following message: "We recently rolled out a system update to Drive item limits to preserve stability and optimize performance. While this impacted only a small number of people, we are rolling back this change as we explore alternate approaches to ensure a great experience for all.".

According to the short post, Google has lifted the file limit. Affected customers should not get the error message anymore when they upload new files to the cloud storage system.

Clearly, the change did affect only a small number of customers. Most Drive users will never come close to the 5 million files cap, but users who have access to multiple Drives may run into the limit, especially considering that Google One has limits of up to 30 terabytes of storage and Google Workspaces Enterprise plans "as much storage" as organizations need.

Google did reveal that shared folders do not count towards the limit, if the files are uploaded by other Google users.

In a follow up tweet, Google Drive promises that it won't make unannounced changes anymore, stating "If we need to make changes, we will communicate them to users in advance". It is unclear if the promise applies to changes related to file limits specifically, or any change that may impact customers on Google Drive.

Closing Words

Google implemented the change without informing customers in advance and without updating documentation or the plans. Google's reversal of the change is clearly a reaction to press coverage, as the company has not reacted in the two months prior to the limits being reported on by various publications.

Now You: do you use online storage?

Google is rolling back the unannounced file limit change on Drive, promises to do better

Researchers Improve ChatGPT By Getting It To Learn From Its Own Mistakes

Tue, 04 Apr 2023 17:04:23 +0000

"It’s not everyday that humans develop novel techniques to achieve state-of-the-art standards using decision-making processes once thought to be unique to human intelligence. But, that’s exactly what we did."

A team of researchers may have found a way of improving large language model (LLM) chatbots, including improving ChatGPT-4's accuracy by around 21 percent. In a new preprint paper, yet to be peer-reviewed, the team explains how they achieved it: allowing artificial intelligence (AI) agents to reflect on their own mistakes.

The team used a process called Reflexion, which "endows an agent with dynamic memory and self-reflection capabilities to enhance its existing reasoning trace and task-specific action choice abilities", according to their paper.

"Human intelligence is notable for its ability to learn from mistakes," the team explained on Substack. "We often don't solve problems on our first try, but when we make mistakes we generate new ideas to refine our approach through self-reflection, through analyzing our missteps."

They tried to replicate this to an extent, by allowing the AI agents to analyze their own actions and mistakes. In the research, AI agents were challenged to solve various problems, from coding to a trial in AlfWorld, a text-based environment that is used to train and test AI agents. In AlfWorld, the agent was asked to complete a number of tasks, but the only way to do so was to learn about its environment through text and be rewarded with observations, like in a text adventure game.

While running the agent in AlfWorld without the reflective technique, it achieved 63 percent accuracy. When the agent was given the ability to reflect on its actions and mistakes, it was able to achieve 97 percent accuracy, solving 130 out of 134 tasks.

In one of these tasks, natural language AI was asked to find the answer to the question "Grown-Ups starred the actor who was best known for which role on 'Allo ’Allo!?" The language model first searched for Grown Ups to view a cast list, and then ’Allo ’Allo! to cross-reference. After failing to get the cast list it needed, it failed the task too.

"I searched the wrong title for the show, ’Allo ’Allo!," the AI explained its reflection process, "which resulted in no results. I should have searched the show’s main character, Gorden Kaye, to find the role he was best known for in the show."

After applying this reflective model, it was given the task again. This time it applied what it learned and finished the task in fewer steps, getting the answer correct.

These AI agents were all powered using ChatGPT-3 and GPT-3.5. In an update, the team used an agent based on ChatGPT-4, and found that when using Reflexion, the AI scored 88 percent accuracy on coding tasks, compared to 67 percent when ChatGPT-4 acted alone.

"It’s not everyday that humans develop novel techniques to achieve state-of-the-art standards using decision-making processes once thought to be unique to human intelligence," the team added on Substack. "But, that’s exactly what we did."

The paper is published on the preprint server arXiv.

Source

Instant Videos Could Represent the Next Leap in A.I. Technology

Tue, 04 Apr 2023 16:20:31 +0000

A start-up in New York is among a group of companies working on systems that can produce short videos based on a few words typed into a computer.

Ian Sansavera, a software architect at a New York start-up called Runway AI, typed a short description of what he wanted to see in a video. “A tranquil river in the forest,” he wrote.

Less than two minutes later, an experimental internet service generated a short video of a tranquil river in a forest. The river’s running water glistened in the sun as it cut between trees and ferns, turned a corner and splashed gently over rocks.

Runway, which plans to open its service to a small group of testers this week, is one of several companies building artificial intelligence technology that will soon let people generate videos simply by typing several words into a box on a computer screen.

They represent the next stage in an industry race — one that includes giants like Microsoft and Google as well as much smaller start-ups — to create new kinds of artificial intelligence systems that some believe could be the next big thing in technology, as important as web browsers or the iPhone.

The new video-generation systems could speed the work of moviemakers and other digital artists, while becoming a new and quick way to create hard-to-detect online misinformation, making it even harder to tell what’s real on the internet.

The systems are examples of what is known as generative A.I., which can instantly create text, images and sounds. Another example is ChatGPT, the online chatbot made by a San Francisco start-up, OpenAI, that stunned the tech industry with its abilities late last year.

Google and Meta, Facebook’s parent company, unveiled the first video-generation systems last year, but did not share them with the public because they were worried that the systems could eventually be used to spread disinformation with newfound speed and efficiency.

But Runway’s chief executive, Cris Valenzuela, said he believed the technology was too important to keep in a research lab, despite its risks.

“This is one of the single most impressive technologies we have built in the last hundred years,” he said. “You need to have people actually using it.”

The ability to edit and manipulate film and video is nothing new, of course. Filmmakers have been doing it for more than a century. In recent years, researchers and digital artists have been using various A.I. technologies and software programs to create and edit videos that are often called deepfake videos.

But systems like the one Runway has created could, in time, replace editing skills with the press of a button.

Runway’s technology generates videos from any short description. To start, you simply type a description much as you would type a quick note.

< https://vp.nyt.com/video/2023/04/02/107416_1_ai-video-2_wg_mobile_720p.mp4 >

AI video generated by the company Runway’s software that depicts “a dog with a cell phone in the park.”

That works best if the scene has some action — but not too much action — something like “a rainy day in the big city” or “a dog with a cellphone in the park.” Hit enter, and the system generates a video in a minute or two.

The technology can reproduce common images, like a cat sleeping on a rug. Or it can combine disparate concepts to generate videos that are strangely amusing, like a cow at a birthday party.

https://vp.nyt.com/video/2023/04/02/107418_1_AI-VIDEO-4_wg_720p.mp4

“A tranquil river in the forest” in an AI video generated by Runway’s AI video software.

The videos are only four seconds long, and the video is choppy and blurry if you look closely. Sometimes, the images are weird, distorted and disturbing. The system has a way of merging animals like dogs and cats with inanimate objects like balls and cellphones. But given the right prompt, it produces videos that show where the technology is headed.

“At this point, if I see a high-resolution video, I am probably going to trust it,” said Phillip Isola, a professor at the Massachusetts Institute of Technology who specializes in A.I. “But that will change pretty quickly.”

Like other generative A.I. technologies, Runaway’s system learns by analyzing digital data — in this case, photos, videos and captions describing what those images contain. By training this kind of technology on increasingly large amounts of data, researchers are confident they can rapidly improve and expand its skills. Soon, experts believe, they will generate professional-looking mini-movies, complete with music and dialogue.

It is difficult to define what the system creates currently. It’s not a photo. It’s not a cartoon. It’s a collection of a lot of pixels blended together to create a realistic video. The company plans to offer its technology with other tools that it believes will speed up the work of professional artists.

Several start-ups, including OpenAI, have released similar technology that can generate still images from short prompts like “photo of a teddy bear riding a skateboard in Times Square.” And the rapid advancement of A.I.-generated photos could suggest where the new video technology is going.

The founders of Runway, Anastasis Germanidis, top, Alejandro Matamala Ortiz, left, and Cris Valenzuela at their Manhattan office.

Credit...Justin J Wee for The New York Times

Last month, social media services were teeming with images of Pope Francis in a white Balenciaga puffer coat — surprisingly trendy attire for an 86-year-old pontiff. But the images were not real. A 31-year-old construction worker from Chicago had created the viral sensation using a popular A.I. tool called Midjourney.

Dr. Isola has spent years building and testing this kind of technology, first as a researcher at the University of California, Berkeley, and at OpenAI, and then as a professor at M.I.T. Still, he was fooled by the sharp, high-resolution but completely fake images of Pope Francis.

“There was a time when people would post deepfakes, and they wouldn’t fool me, because they were so outlandish or not very realistic,” he said. “Now, we can’t take any of the images we see on the internet at face value.”

Midjourney is one of many services that can generate realistic still images from a short prompt. Others include Stable Diffusion and DALL-E, an OpenAI technology that started this wave of photo generators when it was unveiled a year ago.

Midjourney relies on a neural network, which learns its skills by analyzing enormous amounts of data. It looks for patterns as it combs through millions of digital images as well as text captions that describe what each image depicts.

When someone describes an image for the system, it generates a list of features that the image might include. One feature might be the curve at the top of a dog’s ear. Another might be the edge of a cellphone. Then, a second neural network, called a diffusion model, creates the image and generates the pixels needed for the features. It eventually transforms the pixels into a coherent image.

Companies like Runway, which has roughly 40 employees and has raised $95.5 million, are using this technique to generate moving images. By analyzing thousands of videos, their technology can learn to string many still images together in a similarly coherent way.

“A video is just a series of frames — still images — that are combined in a way that gives the illusion of movement,” Mr. Valenzuela said. “The trick lies in training a model that understands the relationship and consistency between each frame.”

Like early versions of tools such as DALL-E and Midjourney, the technology sometimes combines concepts and images in curious ways. If you ask for a teddy bear playing basketball, it might give a kind of mutant stuffed animal with a basketball for a hand. If you ask for a dog with a cellphone in the park, it might give you a cellphone-wielding pup with an oddly human body.

But experts believe they can iron out the flaws as they train their systems on more and more data. They believe the technology will ultimately make video-creation as easy as writing a sentence.

“In the old days, to do anything remotely like this, you had to have a camera. You had to have props. You had to have a location. You had to have permission. You had to have money,” said Susan Bonser, an author and publisher in Pennsylvania who has been experimenting with early incarnations of generative video technology. “You don’t have to have any of that now. You can just sit down and imagine it.”

Source

TSMC To Make Next-Gen Intel Graphics Cards, Release In 2024

Tue, 04 Apr 2023 04:23:37 +0000

The next generation of Intel GPU, named Intel Battlemage, would be made by TSMC on a 4nm process and is expected to release in 2024.

A year ago, Intel announced its first graphics cards in the Intel Arc series. It named that series as Intel Arc Alchemist. While originally it was announced in March, it was released only in October with mediocre reviews.

Most reviewers found that these cards, which include Intel Arc A770, A750 and A380, performed poorly compared to the graphics cards by Nvidia and AMD. They also mentioned how these cards were released a couple or more years later than they should have been. Another thing which was found was that these cards had a lot of problems in running games released years ago.

What Intel did, however, is to massively upgrade its drivers to improve the performance of these graphics cards. Not only that, due to lack of sufficient sales, Intel has also cut the prices of these cards multiple times.

Since then, Intel’s graphics division has seen a lot of changes. This includes structural changes in the leadership department.

There were rumors that Intel might cancel its Arc graphics cards altogether. But thanks to some rational people in Intel, that was not done. Now, more information is coming about the next-generation of Intel Arc graphics cards.

Intel Battlemage Graphics Cards

Intel Arc Battlemage Release Dates Timeline. Credit: Commercial Times.

Taiwan based publication Commercial Times has reported (translation) that Intel’s next-gen Arc Battlemage graphics cards (Xe2-HPG) are going to be made by world’s largest semiconductor maker TSMC. It mentions that while Intel does make its own advanced chips, it is sticking with TSMC to produce the chips for its graphics cards.

While Intel Arc Alchemist was made on a 6nm process by TSMC, Intel’s Arc Battlemage will be made on a 4nm process by them.

The article mentions that the production of these cards are going according to the originally intended speed and Intel’s second-generation, that is Battlemage cards, are all set to start production in first half of 2024. The release of the Intel Arc Battlemage graphics cards is expected to start in the second half of the same year, in 2024.

Intel Celestial Graphics Cards

The above article also talked about the third generation of Intel Arc graphics cards. Named Intel Arc Celestial (Xe3-HPG), it is expected to release in the second half of 2026.

It mentions that Intel Arc Celestial graphics cards will be based on a TSMC’s 3nm N3X process. TSMC will start its production in first half of 2026 and the release is going to happen in the second half of the same year, in 2026.

Conclusion

There’s one thing for sure. More competition is required in the consumer gaming graphics cards. With Intel entering the scene made us all hopeful that the Nvidia and AMD monopoly would be challenged. But Intel Arc Alchemist, while priced competitively now, has been a disappointment.

Here’s hoping that Intel could do better with its next-generation of graphics cards.

These intended timelines could always have delays, though. Intel is known to do that.

TSMC To Make Next-Gen Intel Graphics Cards, Release In 2024

Apple has reportedly started a small number of corporate layoffs

Tue, 04 Apr 2023 04:21:54 +0000

The company has stood out from the big tech crowd in not doing layoffs, but it appears that has come to an end.

Apple is reportedly laying off a small number of people from one of its retail teams, according to reports from Bloomberg and Business Insider. It’s currently not clear how many people will be affected, but Bloomberg says the number is “likely very small,” and both outlets say that, internally, the company is pitching it as a way to improve its operations rather than as a cost cutting measure.

Still, until now, Apple’s lack of layoffs has set it apart from many big tech companies that have announced major cuts. Those include:

Google: 12,000 jobs
Amazon: 27,000 jobs
Meta: 21,000 jobs
Microsoft: 10,000 jobs

Apple’s layoffs seem to be on a massively smaller scale, but it appears it can no longer act as an example of a company that hasn’t resorted to laying off employees.

According to the Monday reports, the jobs Apple’s cutting are in the division that handles building and upkeep for its retail stores, and affected employees have been told that they have until the end of the week to apply for other positions at the company. Apple is offering up to four months of severance pay for those that aren’t able to stay, according to Bloomberg.

While these are the first reported layoffs of full-time employees at Apple since the big tech cuts began, the company has been paring down costs in other ways, with CEO Tim Cook telling The Wall Street Journal that layoffs were “a last resort.” Last month, Bloomberg reported that it’d been laying off contractors, leaving some newly-opened positions unfilled and slowing down hiring for some departments, delaying bonuses, reducing travel budgets, pushing back projects, and more.

Apple has reportedly started a small number of corporate layoffs

There's more cost cutting moves being made at Google including less staplers and tape

Tue, 04 Apr 2023 04:20:52 +0000

After laying off 12,000 employees in January, Google is reportedly making even more cost-cutting moves to reduce spending. That includes reductions in equipment and supplies and some company services.

CNBC reports that Google has informed its remaining workers of these cost-saving efforts via recent company-wide emails and memos. That includes cutting back on staplers and tape. One memo sent to a San Francisco office reportedly stated:

We have been asked to pull all tape/dispensers throughout the building. If you need a stapler or tape, the receptionist desk has them to borrow.

Google already told some employees back in February they were expected to share desks with others when they didn't come into the office. CNBC's new article stated that Google employees who are not working in engineering duties and who need a new laptop would now get a Chromebook by default. Previously, other laptops, including Apple MacBooks, were offered to normal employees.

Google's workers can also no longer expense the cost of a new smartphone to the company if another phone is available for them to use internally. An employee who needs an accessory that costs over $1,000 must now get prior approval, and if one is also not available internally.

Google employee services like food, transportation, and more are also on the chopping block. That's partly due to the fact that many workers now only come into their office three days a week. Google's internal memo says that some of its company cafes could close on Mondays and Fridays, among other measures.

There's more cost cutting moves being made at Google including less staplers and tape

Okay, so ChatGPT just debugged my code. For real.

Mon, 03 Apr 2023 22:13:47 +0000

Not only can ChatGPT write code, it can read code. On one hand, that's very helpful. On the other hand, that's truly terrifying.

So. Programming. For people who program for a living, it's a constant game of mental Jenga: one line of code stacked upon another, building a tower of code you hope is robust enough not to come crashing down.

But it always does. Code never works the first time it's run. And so, one of the key skills for any programmer is debugging -- the art and science of finding why code isn't running or is doing something unexpected or undesirable.

It's a little like being a detective, finding clues, and then finding out what those clues are trying to tell you. It's very frustrating and very satisfying, sometimes at exactly the same time.

I do a lot of debugging. It's not just because code never works the first time it's run. It's also because I use the debugging to tell me how the code is running, and then tweak it along the way.

But while good debugging does require its own special set of skills, it's also ultimately just programming. Once you find out why some block of code isn't working, you have to figure out how to write something that does work.

Real-world ChatGPT testing

This week, I was working on three coding tasks for software I maintain. Two were fixes for bugs reported by users. One was a new piece of code to add a new feature. This was real, run-of-the-mill programming work for me. It was part of my regular work schedule.

I'm telling you that, because up until now, I've tested ChatGPT with test code. I've made up scenarios to see how well ChatGPT would work. This time it was different. I was trying to get real work done, and decided to see if ChatGPT could be a useful tool to get that work done.

It's a different way to look at ChatGPT. Test scenarios are often a bit contrived and simplistic. Real-world coding is actually pulling another customer support ticket off the stack and working through what made the user's experience go south.

So, with that, let's look at those tasks and see how ChatGPT performed.

Rewriting regular expression code

In coding, we have to find a lot of patterns in text. To do so, we use a form of symbolic math called regular expressions. I have been writing regular expressions for decades, and still I dislike doing so. It's tedious, error prone, and arcane.

So when a bug report came in telling me that a part of my code was only allowing integers when it should be allowing for dollars and cents (in other words, some number of digits, possibly followed by a period, and then if there was a period, followed by two more digits), I knew I'd need to use regular expression coding.

Since I find that tedious and annoying, I decided to ask ChatGPT for help. Here's what I asked:

Screenshot by David Gewirtz/ZDNET

And here's the AI's very well-presented reply (click the little square to enlarge):

Screenshot by David Gewirtz/ZDNET

I dropped ChatGPT's code into my function, and it worked. Instead of about 2-4 hours of hair-pulling, it took about five minutes to come up with the prompt and get an answer from ChatGPT.

Reformatting an array

Next up was reformatting an array. I like doing array code, but it's also tedious. So I once again tried ChatGPT. Total failure.

By the time I was done, I probably fed it ten different prompts. Some responses looked promising, but when I tried to run the code, it errored out.

Some code crashed. Some code generated error codes. And some code ran, but didn't do what I wanted.

After about an hour of this, I gave up and went back to my normal technique of digging through Github and StackExchange to see if there were any examples of what I was trying to do, and then writing my own code.

So far, that's one win and one loss for the ChatGPT experience. But now I was going to up the challenge.

Actually finding the error in my code

OK, so this next bit is going to be hard to explain. But think about the fact that if it's hard to explain to you (presumably a human and not one of the 50 or so bots that merely copy and republish my work on scammy, spammy websites), it is even more challenging to explain it to an AI.

I was writing new code. I had a function that took two parameters, and a calling statement that sent two parameters to my code. Functions are little black boxes that perform very specific functions and they are called (asked to do their magic) from lines of code running elsewhere in the program.

The problem was, I kept getting an error message.

The salient part of that message is where it states "1 passed" at one point and "exactly 2 expected" at another. I looked at the calling statement and the function definition and there were two parameters in both places.

W-the-ever-loving-F?

After about fifteen minutes of deep frustration, I decided to throw it to the AI to see if it could help. So, I wrote the following prompt:

Screenshot by David Gewirtz/ZDNET

I showed it the line of code that did the call, I showed it the function itself, and I showed it the handler, a little piece of code that dispatches the called function from a hook in my main program.

Within seconds, ChatGPT responded with this (click the little square to enlarge):

Screenshot by David Gewirtz/ZDNET

Just as it suggested, I updated the fourth parameter of the add_filter() function to 2, and it worked!

ChatGPT took segments of code, analyzed those segments, and provided me with a diagnosis. To be clear, in order for it to make its recommendation, it needed to understand the internals of how WordPress handles hooks (that's what the add_filter function does) and how that functionality translates to the behavior of the calling and executing lines of code.

I have to mark that as incredible, undeniably "living in the future" incredible.

What's it all mean?

As I mentioned earlier, debugging is a bit of art and a bit of science. Most good development environments include powerful debugging tools that let you look at the flow of data through the program as it runs, and this does help when trying to track down bugs.

But when you're stuck, it's often difficult to get help. That's because even a close colleague may not be familiar with the full scope of code you're debugging. The program I'm working on consists of 153,259 lines of code across 563 files -- and as programs go, that's small.

So, if I had wanted to get help from a colleague, I might have had to construct a request almost identically to how I sent it to ChatGPT.

But here's something to consider: I remembered to include the handler line even though I didn't realize that's where the error was. As a test, I also tried asking ChatGPT to diagnose my problem in a prompt where I didn't include the handler line, and it wasn't able to help. So there are very definite limitations to what ChatGPT can do for debugging right now, in 2023.

Essentially, you have to know how to ask the right questions in the right way, and those questions need to be concise enough for ChatGPT to handle the whole thing in one query. That's something that takes actual programming knowledge and experience to know how to do.

Could I have fixed the bug on my own? Of course. I've never had a bug I couldn't fix. But whether it would have taken two hours or two days (plus pizza, profanity, and lots of caffeine) while enduring many interruptions, that's something I don't know. I can tell you ChatGPT fixed it in minutes, saving me a ton of time and frustration.

Looking towards the (possibly dystopian) future

I see a very interesting future, where it will be possible to feed ChatGPT all 153 thousand lines of code and ask it to tell you what to fix. Microsoft (which owns Github) is already working on a "copilot" tool for Github to help programmers build code. Microsoft has also invested billions of dollars in OpenAI, the makers of ChatGPT.

While the service might be limited to Microsoft's own development environments, I can see a future where the AI has access to all the code in Github, and therefore all the code in any project you post to Github.

Given how well ChatGPT identified my error from the code I provided, I can definitely see a future where programmers can simply ask ChatGPT (or a Microsoft-branded equivalent) to find and fix bugs in entire projects.

And here's where I take this conversation to a very dark place.

Imagine that you can ask ChatGPT to look at your Github repository for a given project and have it find and fix bugs. One way could be for it to present each bug it finds to you for approval, so you can make the fixes.

But what about the situation where you ask ChatGPT to just fix the bugs, and you let it do so without bothering to look at all the code yourself? Could it embed something nasty in your code?

And what about the situation where an incredibly capable AI has access to almost all the world's code in Github repositories? What could it hide in all that code? What nefarious evil could that AI do to the world's infrastructure if it can access all our code?

Let's play a simple thought game. What if the AI was given Asimov's first rule as a key instruction. That's a "robot shall not harm a human, or by inaction allow a human to come to harm." Could it not decide that all our infrastructure was causing us harm? By having access to all our code, it could simply decide to save us from ourselves by inserting back doors that allowed it to, say, shut off the power grid, ground planes, and gridlock highways.

I am fully aware the scenario above is hyperbolic and alarmist. But it's also possible. After all, while programmers do look at their code in Github, it's not possible for anyone to look at all the lines in all their code.

As for me, I'm going to try not to think about it too much. I don't want to spend the rest of the 2020s in the fetal position rocking back and forth on the floor. Instead, I'll use ChatGPT to occasionally help me write and debug little routines, keep my head down, and hope future AIs don't kill us all in their effort to "not allow a human to come to harm."

Do you find the fact that ChatGPT can debug helpful or terrifying? Do you think AIs will murder us in our sleep, or do you think we'll be watching our doom with our eyes wide open? Or are you, like me, going to try not to think about it too much because it makes your head hurt? Talk to me in the comments below. While you still can.

Source

‘We are going to the moon’: Jeremy Hansen will be first Canadian astronaut to orbit the moon

Mon, 03 Apr 2023 18:55:05 +0000

On Monday, Hansen was named as part of the four-person crew for NASA’s Artemis II mission, the program’s first crewed mission to travel to the moon since Apollo 17 in 1972.

Jeremy Hansen, a former fighter pilot from London, Ont., will become the first Canadian to orbit the moon.

On Monday, Hansen was named as part of the four-person crew for NASA’s Artemis II mission, the program’s first crewed mission to travel to the moon since Apollo 17 in 1972. The Artemis II is scheduled to launch in November 2024.

It paves the way for the Artemis III mission to return to the surface of the moon targeted for 2025.

The other three members of the crew, mission specialist Christina Koch, pilot Victor Glover and commander Reid Wiseman are all American.

“I am left in awe of being reminded what strong leadership, setting big goals, with a passion to collaborate and a can-do attitude can achieve, and we are going to the moon together,” Hansen said after the announcement. “Let’s go!”

Hansen, 47, was chosen from the Canadian Space Agency’s pool of active astronauts, which also includes David Saint-Jacques, 53, an engineer and astrophysicist from Saint-Lambert, Que.; Joshua Kutryk, 41, a fighter pilot and engineer from Fort Saskatchewan, Alta.; and Jennifer Sidey-Gibbons, 34, a mechanical engineer and combustion scientist from Calgary.

After liftoff from pad 39B at the Kennedy Space Center in Florida, the crew, aboard one of NASA’s Orion spacecraft will circle the Earth twice before heading for the moon.

There it will circle the moon — potentially travelling further away from the earth than any human in history ever has — before returning to Earth for a splashdown in the Pacific Ocean of the U.S. west coast. The exact distance is yet to be determined, but the unmanned Artemis I which circled the moon in 2022, travelled 2.1 million kilometres on its 25-day mission and reached a maximum distance from Earth of 432,210 kilometres.

The Artemis II mission is planned for 10 days.

Ultimately NASA plans to build an Artemis Base Camp on the moon and a Gateway in lunar orbit supporting long-term scientific research and facilitating more human exploration on and around the Moon.

With files from The Canadian Press

Source