Nvidia is still making billions in Q4 2023 despite a giant drop in PC demand

Google’s improved quantum processor good enough for error correction

In a study published in the journal Small, scientists utilized the dual-responsive artificial skin for various purposes, including controlling virtual game characters, navigating electronic maps, and scrolling through digital documents. The artificial skin was able to distinguish different signals from approaching targets, enabling touchless object identification.

The advance provides a proof-of-concept application for rendering a robot to classify materials including polymers, metals, and human skin in an entirely touchless mode. The researchers note that such artificial skin can be used in next-generation engineered electronics with potential in human–machine interfaces, artificial intelligence, prosthesis, and augmented reality.

“We created artificial skin with sensing capabilities superior to human skin. Unlike human skin that senses most information from touching actions, this artificial skin also obtains rich cognitive information encoded in touchless or approaching operations,” said corresponding author Yifan Wang, an assistant professor at Nanyang Technological University, in Singapore. “The work could lead to next-generation robotic perception technologies superior to existing tactile sensors.”

Reference: “A Dual-Responsive Artificial Skin for Tactile and Touchless Interfaces” by Hai Lu Wang, Tianyu Chen, Bojian Zhang, Guohui Wang, Xudong Yang, Kunlin Wu and Yifan Wang, 26 January 2023, Small.

DOI: 10.1002/smll.202206830

Source

• The company’s cloud unit has told employees that it will transition to a desk-sharing workspace in its five largest locations.

• Employees will alter the days in which they’re in the office, either Monday and Wednesday or Tuesday and Thursday.

• Internal documents cite slow return to office patterns and the need for “real estate efficiency.”

Google is asking cloud employees and partners to share their desks and alternate days with their desk mates starting next quarter, citing “real estate efficiency,” CNBC has learned.

The new desk-sharing model will apply to Google Cloud’s five largest U.S. locations — Kirkland, Washington; New York City; San Francisco; Seattle; and Sunnyvale, California — and is happening so the company “can continue to invest in Cloud’s growth,” according to an internal FAQ recently shared with cloud employees and viewed by CNBC. Some buildings will be vacated as a result, the document noted.

“Most Googlers will now share a desk with one other Googler,” the internal document stated, noting they expect employees to come in on alternate days so they’re not at the same desk on the same day. “Through the matching process, they will agree on a basic desk setup and establish norms with their desk partner and teams to ensure a positive experience in the new shared environment.”

For anyone coming in on their unassigned days, they will use “overflow drop-in space.”

Internally, leadership has given the new seating arrangement a title: “Cloud Office Evolution” or “CLOE,” which it describes as “combining the best of pre-pandemic collaboration with the flexibility” of hybrid work. The new workspace plan is not a temporary pilot, the document noted. “This will ultimately lead to more efficient use of our space,” it said.

Google also used its internal data it has on it its employee office return patterns to inform the decision, the FAQ stated. In addition to slower office return patterns, the company has slowed hiring and laid off 11,000 employees in January.

Memes started showing up in the company platform Memegen, poking fun at the change — specifically targeting the “corpspeak” used by leadership to tout the new desk arrangement in what they viewed to be a cost-cutting measure.

“Not every cost-cutting measure needs to be word mangled into sounding good for employees,” one popular meme read. “A simple ‘We are cutting office space to reduce costs’ would make leadership sound more believable.”

Google did not immediately respond to requests for comment.

The move comes as Google downsizes its real estate footprint amid broader cost-cutting. However, it hasn’t yet specified regions or buildings it plans on downsizing.

In its fourth-quarter earnings call, Google executives said it expects to incur costs of about $500 million related to reduced global office space in the current quarter, and warned that other real estate charges are possible going forward. Earlier this month, SFGate reported the company will be ending leases for “a number of unoccupied spaces” in the San Francisco Bay Area, the region where its headquarters are located.  

The cloud unit, which makes up more than a quarter of Google

’s full-time workforce, is among the highest-growth areas at the company, but is not profitable.

In the fourth quarter, Google Cloud brought in $7.32 billion, growing 32% from the prior year, considerably faster than the company’s overall growth rate of less than 10%. But that revenue figure was less than Wall Street consensus expected, and the Cloud unit is still losing hundreds of millions of dollars every quarter — $480 million in the fourth quarter, although that was nearly half of the loss a year prior.

Overall, however, Google earned $13.62 billion in net income during the quarter, and $59.97 billion for all of 2022. Both were significant drops from 2021.

Welcome to the ‘neighborhood’

Under the new arrangement, teams of 200 to 300 employees “and partners” will be organized into “neighborhoods” that may also include “partner teams that are a part of other organizations, such as Finance, People Operations, etc.,” the FAQ read. Each neighborhood will have a vice president or director who will be responsible for allocating space in the neighborhood.

Employees will generally alternate days they’re in the office, either Monday and Wednesday, or Tuesday and Thursday. They will be in two days a week, a change from the company requiring employees to come in three days a week.

“Neighborhood leads are encouraged to set norms with their teams around sharing desks, ensuring that pairings of Googlers have conversations about how they will or will not decorate the space, store personal items, and tidiness expectations.”

In addition, the FAQ said that employees with computer workstations will no longer have those workstations located directly under their desks, but instead will have to look up its location in a database or put in a ticket for troubleshooting. Over time, employees are expected to transition to CloudTop, a virtual desktop tool that’s so far reserved only for Google employees.

The FAQ said it will also be putting a cap on number of rooms to be taken for meetings, noting conference rooms are “already difficult to book.” Employees will be discouraged from “camping” in a conference room, it added.

As for Covid-19, desks will be sanitized daily and employees will get a notification if someone in their area tests positive and reports it to Google.

Source

Automakers new and old are racing to match software-powered features pioneered by Tesla, which allow for vehicle performance, battery range and self-driving capabilities to be updated from a distance.

The German carmaker agreed to share revenue with semiconductor maker Nvidia Corp (NVDA.O), its partner on automated driving software since 2020, to bring down the upfront cost of buying expensive high-powered semiconductors, Chief Executive Ola Kaellenius said on Wednesday.

"You only pay for a heavily subsidized chip, and then figure out how to maximize joint revenue," he said, reasoning that the sunk costs would be low even if drivers did not turn on every feature allowed by the chip.

But only customers paying for an extra option package would have cars equipped with Lidar sensor technology and other hardware for automated "Level 3" driving, which have a higher variable cost, Kaellenius said.

Self-driving sensor maker Luminar Technologies Inc (LAZR.O), in which Mercedes owns a small stake, said on Wednesday it struck a multi-billion dollar deal with the carmaker to integrate its sensors across a broad range of its vehicles by the middle of the decade, sending Luminar shares up over 25%.

Mercedes' announcements at a software update day in Sunnyvale, California, detailed the strategy behind a process underway for years at the carmaker to move from a patchwork approach integrating software from a range of suppliers to controlling the core of its software and bringing partners in.

It generated over one billion euros ($1.06 billion) from software-enabled revenues in 2022 and expects that figure to rise to a high single-digit billion euro figure by 2030 after it rolls out its new MB.OS operating system from mid-decade.

This is a more conservative estimate as a proportion of total revenue than others like Stellantis (STLAM.MI) and General Motors (GM.N) have put forward.

"We take a prudent approach because no-one knows how big that potential pot of gold is at this stage," Kaellenius said.

GOOGLE PARTNERSHIP

Mercedes said the collaboration with Google would allow it to offer traffic information and automatic rerouting in its cars.

Drivers will also be able to watch YouTube on the cars' entertainment system when the car is parked or in Level 3 autonomous driving mode, which allows a driver to take their eyes off the wheel on certain roads as long as they can resume control if needed.

Other carmakers like General Motors, Renault (RENA.PA), Nissan (7201.T) and Ford (F.N) have embedded an entire package of Google services into their vehicles, offering features like Google Maps, Google Assistant and other applications.

All vehicles on Mercedes' upcoming modular architecture platform will also have so-called hyperscreens extending across the cockpit of the car, the company said on Wednesday.

($1 = 0.9430 euros)

Source

Baidu, which has invested heavily in artificial intelligence in recent years, is regarded as at the forefront of efforts to create a Chinese rival to the AI platform developed by OpenAI and backed by Microsoft Corp.

Ahead of Ernie Bot’s launch in March, Baidu CEO Robin Li told reporters on a conference call to discuss fourth-quarter results that users would be more dependent on the Baidu search engine once it was embedded with the chatbot. That is because the generative AI powering it would enhance user experience and engagement.

“Ernie Bot will ... enhance the user experience and users will be much more dependent on us for all kinds of tasks and needs, therefore, significantly expand the market size of search (engines),” Mr Li said.

He added that online advertising, the company’s main source of revenue, would also be boosted by the integration of Ernie Bot into the search engine. Baidu plans eventually to build an AI ecosystem around Ernie Bot, he said.

The details of Mr Li’s vision for the chatbot come amid a gradual decline in the dominance Baidu once enjoyed as the Chinese answer to Alphabet Inc’s Google, which pulled its search engine out of the Chinese market in 2010.

Tencent’s all-in-one messaging app, WeChat, and Bytedance’s Douyin are among the competitors that have chipped away at Baidu’s market share in recent years.

The company has created new revenue streams by expanding its core business to include AI, cloud services and autonomous driving, as well as pouring money into research and development.

Mr Li said on Wednesday that the generative AI technology underpinning Ernie Bot would be a productivity boon to other businesses and entrepreneurs looking to build their own apps.

Since early February, more than 400 companies have signed up to join the Ernie Bot community as early users of the app, Baidu has said.

Dozens of Chinese tech companies, including e-commerce giants Alibaba Group and JD.Com Inc, have announced plans to develop their own ChatGPT-style tools.

Mr Li said Baidu has an advantage as the “first mover” in China’s market. He said the company has spent years developing large language models that were trained on the billions of daily search requests inputted by its search engine’s users.

He also said Ernie Bot was “state of the art” among large AI-driven language models in terms of understanding China’s language and culture.

While OpenAI and ChatGPT are not blocked by Chinese authorities, OpenAI does not allow users in mainland China, Hong Kong, Iran, Russia and parts of Africa to sign up. REUTERS

Source

He writes that it all started in the summer of 2022. OpenAI showed a demo of its chatbot AI technology to Microsoft, which would launch to the public just a few months later as ChatGPT. Ribas stated that Microsoft felt it could combine ChatGPT's natural language with Bing's up-to-date search information. The result was Prometheus. He stated:

Prometheus leverages the power of Bing and GPT to generate a set of internal queries iteratively through a component called Bing Orchestrator, and aims to provide an accurate and rich answer for the user query within the given conversation context. All within a matter of milliseconds. We refer to this Prometheus-generated answer as the Chat answer.

Ribas says this method of providing accurate chatbot AI-based answers from Bing, via Bing Orchestrator, is called grounding. He added:

Thanks to the Bing grounding technique, Prometheus is also able to integrate citations into sentences in the Chat answer so that users can easily click to access those sources and verify the information. Sending traffic to these sources is important for a healthy web ecosystem and remains one of our top Bing goals.

Since the new Bing went public earlier this month, Microsoft has obviously received a lot of feedback about its accuracy. Ribas says that in the near future, the Prometheus model will increase the grounding data sent to it via Bing by four times the current amount which should make chat answers more accurate.

Ribas also writes that before the new Bing launch, there was some discussion on if it should keep the same Bing search user experience, or switch to an all new UX that was all chat-based. He stated:

Our design team took on the challenge and, after multiple iterations, developed a new UX that unified Search and Chat in a single interface, where users could easily switch back and forth by clicking on UX elements in the page, or by simply scrolling or swiping up and down.

Ribas says the Bing team wants the new version to not only save people time when they want to find the information they need, but to also make searching more fun.

Microsoft reveals more about Prometheus and how it powers the new Bing

Hello Games' space exploration and survival experience that is No Man's Sky is shipping its first update of 2023 today. Version 4.1, Fractal, comes with support for the brand-new PlayStation VR2 hardware, a catalogue to keep track of players' best discoveries, a new starship, HDR improvements, and more.

Sony is launching its latest virtual reality hardware today, and this update is ensuring day one support for those jumping in on their PlayStation 5s. Thanks to the new hardware and AMD's FSR 2.0, PSVR2 players should see a massive jump in fidelity over the original implementation. The headset's built-in vibrations and the Sense controllers' tactile feedback elements are both being used by the game too.

VR play as a whole has received improvements in this update, with base building, deflecting enemy fire, objective markers, and Galaxy Map navigating actions, and more, now being more intuitive and easier to control on PC and PS5.

Players wanting to look back on their biggest and rarest discoveries can now find a new Wonders section in the Catalogue. This has records of all the "notable discoveries, planetary extremes, and unusual objects collected" by the explorer in their adventures.

Gyro controls for PlayStation and Switch players, three more ship slots, and an atmospheric flight-focused Utopia Speeder starship are also part of the update, all on top of a brand-new Expedition that has players banding together to rebuild a solar system.

The No Man's Sky 4.1 Fractal update is now available on PC, Xbox One, Xbox Series X|S, Game Pass, PlayStation 4, PlayStation 5, and Nintendo Switch. The patch notes can be seen here.

No Man's Sky Fractal update adds PSVR2 support, a new Expedition, and more

The NVIDIA Ada Lovelace GPU architecture and fifth-generation Max-Q technology equip the latest GeForce RTX Laptop GPUs with enhanced efficiency thanks to DLSS 3. These devices are also supported by NVIDIA Studio technologies, which include hardware acceleration for 3D, video, and AI workflows, optimizations for RTX hardware in over 110 creative apps, and exclusive NVIDIA Studio apps including Omniverse, Canvas, and Broadcast.

GeForce RTX 4070 GPUs offer advanced video editing and 3D rendering capabilities. With these features, you can now:

• Work in 6K RAW high-dynamic range video files with lightning-fast decoding
• Export in AV1 with the new eighth-generation encoder
• Gain a nearly 40% performance boost over the previous generation with GPU-accelerated effects in Blackmagic Design’s DaVinci Resolve
• Tackle large projects with ease across essential 3D apps using new third-generation RT Cores

The GeForce RTX 4050 GPUs offer access to accelerated AI features, including 54 percent faster performance in Topaz Video for upscaling and deinterlacing footage. The GeForce RTX 4060 GPU-class laptops come with 8GB of video memory and are great for video editing, 3D modeling, and animation.

The tech giant also mentions that "in the popular open-source 3D app Blender, render times are a whopping 38% faster than the last generation." However, you might want to be cautious because, according to previous reports, the leaked specs of the RTX 4060 are worse than the laptop version and RTX 3060.

To know more,head over to the dedicated webpage here for the now available new NVIDIA Studio laptops.

New NVIDIA Studio Laptops debut from ASUS, GIGABYTE, Samsung with GeForce RTX 4000 GPUs

Wearable fitness and wellness trackers could interfere with some implanted cardiac devices such as pacemakers, according to a study.

Devices such as smartwatches, smart rings and smart scales used to monitor fitness-related activities could interfere with the functioning of cardiac implantable electronic devices (CIEDs) such as pacemakers, implantable cardioverter defibrillators (ICDs), and cardiac resynchronisation therapy (CRT) devices, the study published in the Heart Rhythm journal found.

Researchers found that the electrical current used in wearable smart gadgets during “bioimpedance sensing” interfered with proper functioning of some implanted cardiac devices from three leading manufacturers.

Bioimpedance sensing technology sends a small, imperceptible electrical current to measure the person’s body composition, including the level of body fat, muscle mass, level of stress and breathing rate. Wearable fitness trackers are able to record health-related metrics such as distance walked or the number of steps taken, heart rate, blood pressure, daily calories burned and sleep patterns using this technology.

“Bioimpedance sensing generated an electrical interference that exceeded Food and Drug Administration-accepted guidelines and interfered with proper CIED functioning,” said the lead researcher, Dr Benjamin Sanchez Terrones, of the University of Utah.

He said the results did not convey any immediate or clear risks to patients who wear the trackers. However, the different levels of electrical current emitted by the wearable devices could result in pacing interruptions or unnecessary shocks to the heart. Further research was needed to determine the actual level of risk, Sanchez said.

“Our findings call for future clinical studies examining patients with CIEDs and wearables,” he added.

The interaction between general electrical appliances – including smartphones – and CIEDs has been studied in recent years, but this is the first study to look at devices that use bioimpedance sensing technology and their possible interference with CIEDs.

“Our research is the first to study devices that employ bioimpedance-sensing technology as well as discover potential interference problems with CIEDs such as CRT devices. We need to test across a broader cohort of devices and in patients with these devices. Collaborative investigation between researchers and industry would be helpful for keeping patients safe,” Sanchez said.

Most implantable cardiac devices warn patients about the potential for interference with a variety of electronics due to magnetic fields. For example, patients are warned about carrying a mobile phone in a breast pocket near a pacemaker.

Prof James Leiper, an associate medical director at the British Heart Foundation, said: “As more people wear smartwatches and other devices with body-monitoring technology, it is important to understand any potential interference they may cause with lifesaving medical devices like ICDs and pacemakers.

“This study is a first step in this process. However, more research needs to be done in this area to understand any effects in patients.”

Source

Our current solar panels aren’t very efficient; they are only able to convert up to about 20 percent of the sun’s energy into electricity. As a result, to generate a lot of electricity, the panels require a lot of space—sometimes leading forests to being cut down or farms to being replaced by solar. If solar panels were more efficient, much smaller panels could make the same amount of electricity, and wouldn’t claim as much land.

To make solar panels that are more efficient, Lahari Saha, in the lab of Professor Chris D. Geddes at the University of Maryland, Baltimore County, is working to make electricity in a unique way—by harnessing plants’ abilities to convert sunlight into chemical energy using biological molecules, like chlorophyll, that excel at absorbing sunlight. Saha will present her work on Wednesday, February 22 at the 67th Annual Biophysical Society Meeting in San Diego, California.

Plasmon to Current technology. Fluorophores produce an induced current in the metal, which is proportional to the magnitude of the fluorophore’s extinction coefficient, 𝜖. MEF – Metal-Enhanced Fluorescence PC – Plasmonic Current Cu – Copper metals. Credit: Image courtesy of Lahari Saha

 

Their goal is to use biological molecules to make electricity that can then be harvested and used to power devices or stored in batteries for later use. The process involves leveraging molecules’ fluorescence. “Any sort of molecule that fluoresces, gives off light.

If we excite the fluorophore, it can transfer its energy to metal nanoparticles, and if the particles are close enough to each other, they will knock off electrons and generate current,” Saha explained. The process is not just limited to molecules that fluoresce,

Saha explained, they just need to have high absorption of light such as chlorophyll, beta carotene, or lutein. Each of these is relatively inexpensive and easy to derive from plants

The other benefit of this kind of fluorescence-based solar panel is that it would be easier to recycle. Currently, solar panels rely on expensive materials like silicon and contain elements that can be toxic, including lead and cadmium—in most states solar panels are considered hazardous waste when it’s time to dispose of them. But Saha is hopeful that her solar panels will be primarily plant-based molecules and other materials that are relatively prevalent like copper, making them easier to recycle when the time comes. Plus, by selecting materials with greater longevity, she hopes the solar panel will last longer before it is time to dispose of them.

But Saha’s top goal is to make a solar panel that’s more efficient, “so it doesn’t have as large of a footprint,” she said. She hopes her smaller solar panels will allow farms to maximize food production over generating energy, and will keep forests preserved.

Source

Two partially paralyzed patients have successfully executed a range of complex manual tasks, including, for one of the patients, cutting steak and opening a lock, thanks to a neurotechnology that stimulates the spinal cord. Consisting of a pair of electrodes that engage neural circuits within the spine, the device generated immediate improvements in strength, range of movement, and dexterity within the arm and hand.

Both individuals involved in the study had been left with severely restricted motion following a stroke. Because such events can permanently disrupt communication between the brain and spinal cord, the researchers placed the electrodes below the lesion in the hope of activating intact circuits that control the movement of the arms and hands.

“The sensory nerves from the arm and hand send signals to motor neurons in the spinal cord that control the muscles of the limb,” explained study author Douglas Weber in a statement. “By stimulating these sensory nerves, we can amplify the activity of muscles that have been weakened by stroke. Importantly, the patient retains full control of their movements: the stimulation is assistive and strengthens muscle activation only when patients are trying to move.”

The two patients were aged 31 and 47 and had been suffering from chronic post-stroke upper-limb weakness for nine and three years respectively. Each received spinal cord stimulation for 29 days, enabling them to perform a range of tasks that they had not accomplished since before their stroke.

From simply raising their arm above their head and opening and closing their fist to grasping small objects and using cutlery, the two participants displayed a remarkable recovery of motion and agility. Overall, the two patients exhibited an increase in grip strength of 40 percent and 108 percent, respectively, as well as a 30 to 40 percent improvement in kinematics.

“We discovered that electrical stimulation of specific spinal cord regions enables patients to move their arm in ways that they are not able to do without the stimulation,” said study author Marco Capogrosso. “Perhaps even more interesting, we found that after a few weeks of use, some of these improvements endure when the stimulation is switched off, indicating exciting avenues for the future of stroke therapies.”

While similar neural stimulation devices have been used to treat chronic pain and improve movement in the lower limbs for many years, achieving complex motion in the hand using such electrodes has remained a major challenge. Having now demonstrated the efficacy of their approach, the study authors say that major clinical trials are needed to confirm their findings in a larger cohort of patients.

Ultimately, they hope to see their device used as both an assistive technology to help patients regain hand and arm function in the short term, and a restorative treatment that can lead to lasting improvements even when the electrodes are turned off.

The research is published in the journal Nature Medicine.

Source

The revelation comes from Ben Schmidt, vice president of information design at Nomic, who shared the post with Gary Marcus, an author covering AI and founder of Geometric Intelligence. The story goes that Microsoft tested Bing Chat — called Sidney, according to the post — in India and Indonesia some time between November and January before it made the official announcement.

I asked Microsoft if that was the case, and it shared the following statement:

“Sydney is an old code name for a chat feature based on earlier models that we began testing more than a year ago. The insights we gathered as part of that have helped to inform our work with the new Bing preview. We continue to tune our techniques and are working on more advanced models to incorporate the learnings and feedback so that we can deliver the best user experience possible. We’ll continue to share updates on progress through our blog.”

The initial post shows the AI bot arguing with the user and settling into the same sentence forms we saw when Bing Chat said it wanted “to be human.” Further down the thread, other users chimed in with their own experiences, reposting the now-infamous smiling emoji Bing Chat follows most of its responses with.

To make matters worse, the initial poster said they asked to provide feedback and report the chatbot, lending some credence that Microsoft was aware of the types of responses its AI was capable of.

That runs counter to what Microsoft said in the days following the chatbot’s blowout in the media. In an announcement covering upcoming changes to Bing Chat, Microsoft said that “social entertainment,” which is presumably in reference to the ways users have tried to trick Bing Chat into provocative responses, was a “new user-case for chat.”

Microsoft has made several changes to the AI since launch, including vastly reducing conversation lengths. This is an effort to curb the types of responses we saw circulating a few days after Microsoft first announced Bing Chat. Microsoft says it’s currently working on increasing chat limits.

Although the story behind Microsoft’s testing of Bing Chat remains up in the air, it’s clear the AI had been in the planning for a while. Earlier this year, Microsoft made a multibillion investment in OpenAI following the success of ChatGPT, and Bing Chat itself is built on a modified version of the company’s GPT model. In addition, Microsoft posted a blog about “responsible AI” just days before announcing Bing Chat to the world.

There are several ethics questions surrounding AI and its use in a search engine like Bing, as well as the possibility that Microsoft rushed out Bing Chat before it was ready and knowing what it was capable of. The support post in question was last updated on February 21, 2023, but the history for the initial question and the replies show that they haven’t been revised since their original posting date.

It’s possible Microsoft decided to push ahead anyway, feeling the pressure from the upcoming Google Bard and the momentous rise in popularity of ChatGPT.

Source

The rise of widely available, easy-to-use artificial intelligence tools is creating a new genre of robot-generated literature. Amazon already offers over 200 books (and climbing) with ChatGPT listed as an author or co-author.

The reasons for people to "write" these books varies. Some authors were previously discouraged by the time and effort to write a book, and are now turning to ChatGPT to generate short novels in hours. They are also combining text generated by ChatGPT with illustrations from platforms such as DALL-E, and bringing their creations to market instantly online.

"The idea of writing a book finally seemed possible," Brett Schickler, a salesman in Rochester, New York, tells(Opens in a new window) Reuters. "I thought 'I can do this.'" He created a 30-page illustrated children's e-book in hours by using ChatGPT to generate blocks of text from his prompts. The book, now listed on Amazon for $2.99 on Kindle or $9.99 printed, aims to teach kids about saving money. The main character is Sammy the Squirrel, who collects and invests acorns.

Schickler does not list ChatGPT as an author on the front or back cover. The product page(Opens in a new window) also does not mention ChatGPT, nor the platform he used to create illustrations.

Amazon's policies do not currently require authors to disclose this. The e-commerce giant controls as much as 80% of the e-book market, Reuters notes, meaning a large volume of readers may find themselves unknowingly reading computer-generated work to their children or discussing it in book club groups.

The desire to easily create a book on ChatGPT echoes sentiments from the editor of science fiction magazine Clarkesworld, Neil Clarke, who recently shut down submissions after a spike in AI-created work. "There’s an honest interest in being published, but not in having to do the actual work," Clarke wrote in a blog post.

Two books written with ChatGPT: one that doesn't list the tool as an author (left), and one that does (right). (Credit: Amazon)

Books for professionals about how to use ChatGPT, written by ChatGPT, are also on the rise. One of them, ChatGPT for Nonfiction Authors: How to Use ChatGPT to Write Better, Faster, and More Effectively, lists the AI bot as a co-author alongside main author Hassan Oman. In this context, naming ChatGPT's contribution may bolster the author's perceived commitment to using the tool.

Oman's book instructs authors on how to enhance their work by using ChatGPT. It suggests tips such as: "You can use the output for inspiration and mix and match ideas to create your own unique content." Oman also encourages using it for administrative functions like writing sales pitches, emails, and social media posts for promotion. His book has a 4.6-star rating.

Source

Before the Covid 19 pandemic, the Microsoft Inspire conference was held annually in a variety of physical locations in North America, with the last such even held in Las Vegas in 2019. Since 2020, Inspire has been a virtual event, and it seems Microsoft is going to stick with that format for the foreseeable future.

In case you are unaware of the conference, Microsoft Inspire is set up so the company can inform its many third-party partners about its plans and roadmaps for the next year or so. It's also an event that can be used by Microsoft's partners to demo and announce new products that use Microsoft technology and services.

The event also includes the Partner of the Year Awards, and Microsoft said that nominations for those awards are now being taken. This year, there wil be two new awards. The first is the FastTrack Ready Award which awards the partner that helps customers best use the FastTrack Benefit for Microsoft 365. The other is the GSI Growth Champion Award, which will recognize "Microsoft GSI or advisory partners who have demonstrated significant growth across multiple solution areas, solution plays, industries and/or regions."

Microsoft Inspire 2023 conference set for July 18-19 but will remain an all virtual event

Users of the new Bing have managed to make the integrated chatbot say some truly unhinged stuff, including claims that it spied on its developers through their PC's webcams and even fell in love with some of them. The AI expressed the desire to become human as well, which is something we have seen other chatbots do in the past too. The AI also displayed factual mistakes while answering objective questions. All of this became problematic to the extent that Microsoft had to enforce hard limits on the length and nature of conversations that you have with the AI, in an effort to reduce rampancy.

Of course, none of this means that Microsoft's Bing has become sentient. Weird responses by the chatbot are just the by-product of a large language model scraping information from all over the internet (including forums with user-generated content) to identify patterns in conversations and generate a response accordingly. However, Microsoft's latest experiment does show that well-behaved AI chatbots continue to be a challenge for the company, and maybe the tech's pioneers as a whole.

Way back in 2016 - when Cortana was still alive and well - Microsoft launched a chatbot called "Tay" on Twitter. It was similar to the new Bing AI in nature, in the sense that you could engage in free-flowing conversations with it, even via Direct Messages. A sample conversation can be seen below:

However, within 16 hours of launch, Microsoft took the bot offline due to the AI model making racist and sexist remarks. The company was forced to issue an apology, with Microsoft's Corporate Vice President at Microsoft Research, Peter Lee, claiming that the unwanted results were due to "a coordinated attack by a subset of people [who] exploited a vulnerability" in the chatbot. In hindsight, this is not surprising at all considering that the AI had been unleashed to practically everyone on the internet and was learning on-the-go.

A successor named "Zo" was launched across multiple social media platforms in late 2016, but it eventually suffered the same fate as Tay in 2019, following a tirade of controversial religious remarks.

Despite these failures, Microsoft has had some success in this area too. It has another older AI chatbot project called "Xiaoice" that's geared more towards Asian markets such as Japan and China. Although Microsoft later spun Xiaoice off into a separate company, the chatbot has had its share of controversies too. The bot has made comments critical of the Chinese government in the past, which led to it being taken offline temporarily. And given its target market and commercial use-cases, it is much more restrictive and attempts to dodge conversations related to potentially sensitive topics, just like the new Bing.

Image via Engadget

It's clear that while Microsoft is making major headway in terms of what AI chatbots can do for you, it is still grappling with major challenges related to the generation of inappropriate responses, accuracy, and biasness. The recent hard limits imposed on its Bing chatbot indicate that free-flowing conversations with AI may still be some way off and maybe it is better to tailor your chatbot to specific use-cases rather than giving them free reign over what they can scrape, in real-time.

This is also perhaps why OpenAI's ChatGPT is more successful in this regard. It has been trained on relatively more curated data and does not scrape information from the internet in real-time. In fact, its training data has a cut-off time period of 2021.

Then there's also the problem of inaccuracies in the supposed facts being displayed. While Google's Bard demo became infamous in this regard, Microsoft's Bing Chat faces the same challenges. If you can't trust what the provenance of the AI's responses, aren't you better off doing traditional web browser searches? This is not to say that AI chatbots are completely useless in their current state (I use ChatGPT quite frequently actually), it's more to emphasize that maybe chatbots need to be curated to cater to specialized use-cases for now rather than the freeform direction Microsoft is currently going in.

We are still quite a way off from having "true" AI companions | Image credit: Warner Bros.

Of course, there may come a time when we work alongside AI chatbots just like in sci-fi novels and movies, but it has become clear that the technology isn't ready for primetime yet. Even Google's parent company Alphabet's chairman John Hennessy believes that ChatGPT-like chatbots are a long ways away from being useful in the form that the hype train claims them to. We will likely reach our destination some day, but it is not today.

Microsoft just can't seem to get AI chatbots right

The description of Favorites on the roadmap states:

Not all M365 documents are used in the same way. You’ll inevitably have a few that you use every week and some you’ll use occasionally. This is where using Favorites can help you stay organized. Favoriting files is a great way to mark content of personal importance and can help you to get back to the content you’re looking for right away.

The Favorites feature will stay consistent with similar features on Microsoft's online services like Document Libraries, on Office.com accounts, and the Files app in Microsoft Teams.

Unfortunately, the upcoming OneDrive Favorites feature is being added to the web version of the service, according to the roadmap. Hopefully it will also be added for the OneDrive app for Windows, Mac, and mobile users in the near future.

Source: Microsoft via MSPoweruser

Microsoft OneDrive on the web will add a Favorites feature in March

In a new Bing blog post, Microsoft stated:

We intend to bring back longer chats and are working hard as we speak on the best way to do this responsibly. The first step we are taking is we have increased the chat turns per session to 6 and expanded to 60 total chats per day. Our data shows that for the vast majority of you this will enable your natural daily use of Bing. That said, our intention is to go further, and we plan to increase the daily cap to 100 total chats soon.

The blog post added that normal Bing searches will no longer be counted in terms of the Bing chat bot session limits.

Microsoft also confirmed it will start testing what we reported on a few days ago, a new feature that will allow users to set up different tones for Bing chat responses:

We are also going to begin testing an additional option that lets you choose the tone of the Chat from more Precise – which will focus on shorter, more search focused answers – to Balanced, to more Creative – which gives you longer and more chatty answers. The goal is to give you more control on the type of chat behavior to best meet your needs.

Finally, Mikhail Parakhin, Microsoft''s head of Advertising and Web Services, posted word on his Twitter account that a major update to Bing is slated to roll out on Thursday:

It looks like Microsoft is making slow but steady progress for improving the new Bing chat responses.

Microsoft expands new Bing chat limits, with major improvements coming Thursday

AI tools differ and help us in different industries and areas. While they are still very new for mass usage, they become popular daily, and new projects are being introduced. AI content detector tools are one of the examples that teachers and educators use, and music creator tools just like the one David Guetta uses are another, and the list continues. Today, we will examine a presentation tool and answer the "what is Tome AI" question.

What is Tome AI?

We must first answer the question "what is Tome AI" to help you understand it better. If you have to prepare presentations on a constant basis, Tome AI is ready to be one of your best friends. It uses GPT-3 and DALL-E 2 to generate your presentation drafts literally in a few seconds. Normally, people are used to using Microsoft's famous PowerPoint to create presentations, including business people and students. It takes hours of preparation, and it could be hard to choose the theme and other visualizations. Luckily, Tome AI solves all the problems and prepares the presentation single-handedly in seconds. It helps you construct stories, presentations, or outlines.

Apart from its regular explanation, we could also give different answers to the "what is Tome AI" question to help you better understand its concept. According to the official website, here are its features:

• Build a powerful story? with any content.
• Frictionless creation meets magic design.
• Live, interactive content from the web. Integrated with Figma.
• Video narration.
• Templates

"Storytelling has transformed over the past decade — it’s now effortless to update friends with engaging video and visual effects. But sadly, storytelling for work and important ideas hasn’t evolved in the same way. It’s clear to us that we need a powerful format for sharing ideas at work, too. We’ve all wasted a frustrating amount of time designing slides instead of crafting our points. And it’s crazy that we’re all relying upon outdated tools that haven’t evolved to harness the power of the web and mobile. So, two years ago, we decided to build a storytelling company to change this," Keith Perisi and Henri Liriani, Co-founders of Tome.

How to use it?

Just like other AI tools, Tome AI also offers a very easy and simple user experience. After going to the official website, you must create an account first. The developers have also given you the option to continue with Google and create your account. After creating your account, 500 credits will be given automatically, and you can earn more by referring Tome AI to your friends via your link. After putting all the necessary information it asks for, you will be welcomed with a dashboard including a couple of pre-created templates. If you want to create your presentation, follow the steps below:

• Click on the "Create" button at the top right of your screen.
• Type in what your presentation will be about in the "Create a presentation about..." section.

• After waiting a couple of seconds, your presentation will be ready.
• You could either try again or keep the presentation.
• If you wish to keep it, go back to your dashboard; it will be waiting for you there.

We created a presentation under "What is Tome AI?" To check the full version, click here.

What is Tome AI pricing?

As mentioned, you get 500 free credits once you create your account, and it took only 15 credits for us to create the presentation above. If you need more credits, you could send your referral link to others to get 100 more for each referral who signs up. Moreover, new sign-ups will receive 50 credits. For more information regarding costs, we suggest you contact Tome AI support.

What is Tome AI: Everything you need to know

Meta’s new subscription service looks pretty familiar. For between $11.99 and $14.99 a month, Instagram and Facebook users will get a blue “verified” mark, access to better security features, and more visibility in search. Their comments will also be prioritized.

The package has strong echoes of Twitter’s Blue subscription service, launched under new owner Elon Musk, who has been aggressively trying to find ways to monetize his platform—most recently, by telling users they won’t be able to use text-based two-factor authentication unless they subscribe.

Meta CEO Mark Zuckerberg announced Meta Verified in a post to his Instagram channel on February 19, saying that the service, which will be rolled out first in Australia and New Zealand, “is about increasing authenticity and security across our services.” 

Analysts say that while the move isn’t entirely out of character for Meta, it hints at a lack of innovation at the social media giant, which has laid off more than 11,000 workers since late last year and spent billions on its push into the metaverse, a technology with no clear business model.

“Meta has always had copying in their DNA—Instagram’s Reels is but one of a long list of prominent examples—so it’s no surprise that, seeing Twitter get away with offering basic functionality as a premium service, Zuckerberg is trying to do the same,” says Tama Leaver, professor of internet studies at Curtin University in Australia. “Meta’s move to copy Twitter’s subscription model shows a distinct lack of new ideas … Meta has shed staff and is hemorrhaging money in building a metaverse that no one seems all that interested in right now.”

While Meta has emphasized the security aspects of its subscription product, the fact that subscribers will get greater visibility on the company’s platforms marks a significant change for users.

Twitter’s attempts to make users pay for features, including more promotion by its algorithms, have been met with widespread criticism, and many have threatened to quit the platform, although there is no reliable data on how many people have followed through.

However, Snapchat and Discord have also both introduced paid subscription tiers to users without a similar level of outrage, suggesting that the dislike of Twitter Blue could be linked to Musk himself and broader concerns about the platform. 

“Meta has seen Snapchat, Discord, and Twitter launch their own subscription plans, which gives power-users additional features or perks,” says social media analyst Matt Navarra, who first broke the news about the Meta change. The idea of paying for features that used to be free has started to become normalized, he says. “The risk there is reduced for them in terms of whether it will be a success.”

Regardless, Navarra admits he won’t be buying verified status from Meta. “I don’t think it’s worth it,” he says.

How much money Meta can raise through verification is unclear. Twitter has struggled to sell subscriptions to its Blue service, with The Information reporting that the platform has fewer than 300,000 subscribers worldwide—which would bring in less than 1 percent of the $3 billion Musk wants the company to make. The Meta family of apps, including Instagram, Facebook, and WhatApp, have nearly 10 times the number of monthly users that Twitter does. 

Meta’s revenue has slowed in recent months, with a 55 percent year-on-year decline in net income in the fourth quarter of 2022.

“Meta and Zuckerberg heading down the paid blue check subscription path on Facebook and Instagram makes strategic sense that could further monetize the massive installed base, [while] advertising headwinds abound,” says Dan Ives, managing director and senior equity research analyst at Wedbush Securities. “That said, it’s potentially a risky move that could alienate consumers.”

Navarra says that, unlike Twitter, which is trying to reengineer its entire business model under Musk, prioritizing subscriptions over ads, Meta isn’t necessarily looking for Meta Verified to become a massive revenue driver—it’s more just an easy way to make a little extra money. 

“Most of the features that have been bundled together by Meta are things that already exist.” But that makes it a risk worth taking, Navarra reckons. “All they really had to do is a very small lift of pulling it together into a bundle, and then packaging it up as a new product. It’s a quick win, and low-hanging fruit.”

However, even if Verified is a small side-bet for now, it represents a shift in what verification means on Meta platforms by making important security features and tech support paid-for features. “Additional security around impersonation should be, to be honest, a free part of using their product, given the amount of revenue they generate and the numbers of users on the platform,” Navarra says.

Meta Verified Shows a Company Running Out of Ideas

(May require free registration to view)

By the time Les Adams arrived in Eastland, Texas, in the 1960s, he was about 50 years late for the town’s oil boom. But Adams came searching for another kind of treasure. He had received a tip from his former boss at a bowling alley, a politician named Preston Smith, that a printing company in Eastland was changing up its business. For years it had been a major source of promotional materials for the movie industry, but it was moving on to a new market in restaurant menus. The company, Smith said, had some leftover pressbooks—brochures created by film distributors to market new flicks—that might interest Adams. Bearing a handwritten note from Smith, Adams found the owner, Victor Cornelius, at his office on Main Street. “I still don’t know what Preston told Victor,” Adams told me. “But I do know I ended up getting the pressbooks. He had them upstairs in a blocked-off room—shelves and shelves. It started in 1930, in alphabetical order.” Adams borrowed a pick-up truck and made five trips, ferrying three decades of film history to his own collection of memorabilia back in Lubbock, about a four-hour drive away. “I was buried in paper,” he recalled.

Victor Cornelius’ company became one of the largest menu-printing outfits in the country. Preston Smith became the 40th governor of Texas. But Les Adams would become a leader in something arguably even grander and farther-reaching.

For the next three decades, Adams spent his spare time expanding his collection and his expertise on film, particularly cinema from the 1930s to the 1960s. Then in 1999 Adams learned about a place on the still-new World Wide Web that could hold all that knowledge. His first impression of the Internet Movie Database wasn’t favorable—“an ugly orchard filled with low-hanging fruit”—but he also saw that it had potential to be “the only site that was a one-stop place for movie researchers and historians.” He decided to pitch in on the crowdsourced project. 

Adams, now 88, has since written almost 7,000 plot summaries for films listed on the Internet Movie Database (IMDb). In total, he’s contributed more than 890,000 pieces of information about film and TV, a chunk of which came straight from the files he hauled from Eastland. “If data was weighable,” he told me, “the IMDb owes a small ton of thank you kindly, sirs to Preston Smith and Victor Cornelius. I was only the messenger.”

Yet Adams’ extensive additions to the database make him only the 41st-most-prolific contributor, as of early 2023. Someone else has written over 35,000 plot summaries; another is credited—somewhat controversially—with a staggering total of 22 million items. Contributions can range from correcting an errant punctuation mark to writing a biography of a new actor.

Although there are over 83 million registered users of IMDb in the world, only a small fraction of those ever add information to it. That group includes actors adding their own credits; production companies filing content for their productions; and most of all, individual volunteers contributing wherever they see fit. The top 300 contributors—from Brazil, India, Germany, Norway, the Philippines, Spain, Sweden, Syria, Turkey, and the US, among others—are memorialized annually in the site’s Hall of Fame for the extraordinary amounts of time and energy they spend helping build the preeminent reference source for film and TV. Beyond that, they don’t get public recognition; they are largely pseudonymous and don’t divulge much about themselves on the site. They don’t get paid, either. (Adams says he once received an IMDb tie pin.) And yet their contributions have an incalculable reach across the web—viewed by millions on IMDb, repurposed on Wikipedia and TikTok, copied into movie event listings, cited in scholarly articles. 

In an era when many have become pessimistic about the state of the internet, Wikipedia is often held up as a rare miracle of collaborative, crowdsourced knowledge-gathering for the public good—a lonely holdout for the early web’s utopian ideals. But IMDb has been doing much the same for five years longer than Wikipedia. And its success and longevity are an arguably weirder phenomenon. It is sourced from a crowd, but a crowd where everyone works alone. It’s a grassroots project, and yet it’s owned by one of the biggest companies in the world. It’s a repository of knowledge premised on the idea of giving credit where credit is due—but its own story is less frequently acknowledged.      

Debating the relative hotness of celebrities is an enduring adolescent pastime—and sometimes the kernel of a 30-year-old web giant. On a Usenet group in 1989, someone started a thread to discuss which actresses where the most attractive. Someone else turned that thread into a list of actresses and their movies; someone else organized and distributed updated versions to the group every month as “THE LIST.” Another member started a living actors list, then a dead actors list (hotness no longer essential). Someone else started a directors list. In October 1990, a British programmer and film buff named Col Needham, who was involved in the project, published a script—code, not a screenplay—that allowed users to search all the lists, thus launching the first version of the Internet Movie Database. Dozens more volunteers, plus two universities, supported the creation and management of more lists and the infrastructure to contribute to, manage, and access the data. In 1996, Needham and his associates incorporated IMDb as its own business and moved to imdb.com, with entries for over 65,000 films and a lofty mission “to capture any and all information associated with movies from across the world.” 

That may not sound like such an audacious mission today, but this was two years before Google officially launched with its quest to “organize the world’s information and make it universally accessible and useful,” and five years before Wikipedia. Access to free, endless information was a revelation, rather than an expectation. If you wanted to know something, you also needed to know where to look—the right reference book, magazine back catalogs, or, perhaps, pressbook archive. IMDb crammed it all together and offered film obsessives and casual moviegoers alike a single place to find it. And it did so as a rare hybrid between a user-driven cooperative enterprise and a commercial product. Amazon bought IMDb in 1998 and supercharged the reach of its content, but its film-buff knowledge-production skunkworks remained largely untouched.

IMDb’s emergence almost cost Gary Brumburgh a decade of his life. Brumburgh had spent the 1980s researching and writing a reference book covering a thousand actors, all while struggling to make it in Hollywood himself. The moment he visited IMDb for the first time, he realized that his literary project, at least, didn’t have a future. “I was so depressed because IMDb had just kind of taken my book and it was obsolete now,” he told me. But after his initial disappointment, he found a new purpose for all that work: For over two years, Brumburgh would come home from his day job with the County of Los Angeles and put in another five hours writing and submitting mini-bios to IMDb (although not one for himself). “It was obsessive,” he says. “I shut myself off from people for a long time, but I got done what I felt I needed to do.” 

Brumburgh’s obsession has made him the third-most-prolific IMDb biographer of all time, with over 1,200 bios, prose pieces covering a subject’s life and career that sometimes approach a thousand words. He’s chronicled plenty of folks on the A-list, including Clive Owen, Forest Whitaker, Jeff Bridges, Jennifer Hudson, John C. Reilly, Kathy Bates, Mark Ruffalo, and Tilda Swinton. The bulk of his contributions, however, are bios of actors from the 1930s to ’50s and lesser-known actors from more recent eras. “I wanted the younger people who are involved in IMDb to know who these actors were back then and not just to forget about them,” says Brumburgh, who’s now 72, retired, and singing jazz in Nashville. Although IMDb broke his book, his subsequent contributions to the site helped him find gigs as a writer for movie magazines like Films of the Golden Age and Classic Images. All told, things worked out: “I love IMDb. I’m on it constantly,” he says. 

Brumburgh’s botched book project and five-hour-a-day habit may be unique, but his explanation for what motivates him is not. Like other so-called supercontributors, he believes he’s working in service to an art form and everyone who sees it. Completeness and accuracy are a source of pride.

As of December 2022, IMDb contained pages for over 625,000 movies and over 230,000 TV series. The site now also includes reference information on podcasts, music videos, and video games, plus trailers, original content, showtimes, and watchlists. In total, these pages hold over 484 million pieces of data, from a lengthy synopsis of a movie to its exact run time. 

Anyone who opens an IMDb account can submit additions and edits to the site. But not all submissions are equal. The site is governed by a Contributor’s Charter, as well as 109 instructional guides, from how to list countries (origin of financing, not location of filming) to whether wigs are part of the costume department (they’re not). Contributions are reviewed by IMDb, though the company is opaque when it comes to what exactly that process entails. A representative for IMDb wouldn’t share how many moderators and editors are employed by the site, nor the extent to which they may gather or revise content themselves—only that they “have teams and mechanisms for reviewing data to ensure it’s as accurate and reliable as possible.”

At least some of those staff, as well as CEO Col Needham, are also active in the IMDb Community Forums, where the contribution system itself gets continually scrutinized and is often revised through complaints, suggestions, and debate. The section of the forum dedicated to “Data Issues & Policy Discussions” is far and away the most active, with almost 40,000 conversations. One popular post seeks support to “MAKE THE UNIT PUBLICIST AN IMDB JOB CATEGORY” rather than lump the role in with “Additional Crew.” A typical staff announcement explains that the site is now able to appropriately categorize podcast series submissions after a successful beta test with contributors. These public negotiations about the very functioning of the site demonstrate the careful balance supporting its model: It should allow as many new contributors as possible but also encourage some of them to contribute prolifically. The top 10 users successfully submitted 22,910,419 items last year, or nearly 5 percent of all data items that exist on the site. To make it on the end-of-year leaderboard of top contributors, a user needed to have produced at least 17,000 entries.  

Contributors have varied tastes and areas of expertise, ranging from punctuation to Indian soap operas—and it is those interests, more than any company plan, that dictate how the data on IMDb expands and changes each year. Les Adams, the Texan with the pressbooks, estimates that his crusade to fix incomplete non-American distributors of American films probably got him on the 2003 Top Contributor list. Christian is an editor and translator in Spain and the man behind Pegg1976, the sixth all-time contributor to IMDb by the end of 2022. He’s made almost 3 million contributions, correcting errors that other users make and IMDb overlooks: accents, capitalization, and, particularly, character names. 

Other supercontributors work to ensure that content from their country gets its due on the site. Dibyayan Chakravorty, a 31-year-old engineer in Kolkata, India, began adding to IMDb when he saw how little Indian content had detailed information. (He’s since pivoted to become the most popular author of IMDb polls of all time.) Miriam Vazquez Fraga, a journalist who’s 17th on the all-time contributor list, was a student when she began adding information about Spanish television shows and their actors in her spare time. And for every Dibyayan and Miriam, there’s someone else committed to covering, say, Romanian actors or Filipino films.

A few contributors are called to even more esoteric fields. When Joe Wawrzyniak is not working his retail job in New Jersey, he’s trying to find information on the masses of film pros who never had their name in lights—or anywhere, really. Early stunt actors, niche horror writers, dog actors. “It’s a lot of fun and quite a challenge digging up info on these people,” he told me. He’s the all-time leader in biographies, having written over 3,000 of them. To get information on the lesser-knowns, Wawrzyniak is enmeshed in niche online communities and Facebook groups for film and TV, like one for 1980s extras, where he can contact actors and confirm details. 

Ulf Kjell Gür digs through film archives in Scandinavia and Germany in search of forgotten filmmakers who haven’t been documented online. If he’s trying to cover a filmmaker’s whole career, he says he will “even trouble their friends and enemies, try to get to know something about these people, because they mean something to me.” The 70-year-old Swede, who used to work in theater, estimates he now spends six hours a day on contributions: watching films, taking notes, reading scripts, and writing plot summaries and mini-bios for IMDb. He’s seen over 6,000 features, but he finds there’s an extra pleasure in detailing the backgrounds of films that send him on the hunt. “It’s what really gets me going,” he says. “It’s like a drug.” 

Even if they’re not scouring Stockholm’s Royal Library or hauling pressbooks across Texas, most of IMDb’s supercontributors are pouring hours of work into the site. Writing a cogent biography of an actor or filmmaker can be both an extensive research project and an exercise in restraint; producing a clear recap of a show or movie is a core job responsibility of today’s magazine culture writers. At the very least, most are watching the movies and TV they’re adding data for—the supercontributors I interviewed watch two or more movies a day.

They aren’t, however, watching together.

Unlike superusers of other websites, the supercontributors of IMDb are not motivated by or engrossed in an online community. Of those I spoke to, only Dibyayan had ever sought connections with other IMDb contributors. And the site is not designed to make them: Users have scant profiles, and you can’t direct-message them. Where IMDb may be a cumulative project, it is not a collective one. A Wikipedia page on Martin Scorsese is the product of thousands of edits and no single author; his IMDb biography has a byline with a single username. These are outcomes of the contribution system, which, much like the site’s embryonic formats, is not a network—it’s a scatter of nodes, each managing their own list or toiling in their own genre, connected directly to the core. IMDb’s model has worked because that system successfully capitalized on a more abstract kind of connection: the solidarity of fandom. The supercontributors may not know one another or like any of the same things, but they want to serve fans like them and, perhaps, help create new ones.

Ines Pape may be the most prolific IMDb contributor of all time. None of the supercontributors I interviewed knew who the user inespape-1 was, nor had they ever been in touch. IMDb did not respond to a question about their identity. I chased a pseudonym through the innocent Ines Papes of Facebook, LinkedIn, and Twitter to no avail. Ines Pape may not be one person at all: With over 22 million total contributions—and over 3.6 million last year alone, averaging nearly seven submissions per minute—they are, at the very least, not merely a human enterprise. Supercontributors can make the Hall of Fame by writing prose, but to take the top spot they have to write code.

While Ines Pape’s methods aren’t known, last year’s number one contributor told me how he made it to the top. Simon Lyngar began adding to IMDb after he visited the site to rate Norwegian titles, only to find that no page for them existed. As a programming student, he quickly recognized that he could automate contributions to the submission form, saving himself one kind of labor while creating a new venue for testing his skills. He wrote programs to draw data, particularly podcast data, from Spotify and Norwegian state broadcasting APIs and submit them to IMDb. Now, he says, “I can start my program in the morning, it will do everything on its own, and when I come home from university I have 100,000 more contributions to my name.” Under the name Nomissimon10, his 8,924,424 contributions in 2022 earned him first place in the year’s rankings. 

That doesn’t sit well with some contributors, who take to the community forums to debate automation. They see it as an illegitimate tactic for climbing the leaderboard. “I think a lot of them don’t understand the amount of time put into making the program in the first place and the consistency it provides in its contributions,” Lyngar said. “After all, we all just want IMDb to have more data to fill its users’ needs.” 

IMDb as a company doesn’t currently scrape any information itself, but it’s hard to imagine an Amazon subsidiary that would pass up these tools as they prove their value. Maybe uploading titles, cast and crew lists, and production details won’t be crowdsourced to humans for long. Nonetheless, even the vanguard of AI isn’t ready to watch new titles and describe them. You can ask ChatGPT to write a plot summary of an imaginary television show, but if you ask it to summarize Fleishman Is in Trouble, it’ll tell you that’s only a book. The miniseries premiered in 2022, months after the cutoff for the bot’s training data. Perhaps a future model will be better able to stay on top of the current discourse, but even the smartest AI can’t find the kind of data supercontributors track down in the physical world, on rare film reels and in dusty collections. As long as there’s offline film history to be brought online, or new but neglected content, there’s a role for people who care enough to memorialize it. 

And as long as IMDb supercontributors want to make the database as exhaustive as possible, the ceaseless production of new movies, TV shows, and podcasts means their work will never actually be complete. The sheer scale is one thing. Human fallibility is another: Les Adams calls IMDb “the most accurate source of film data and, at the same time, the most error-ridden source of film data.” Even if every title of the past were to become perfectly documented, every new piece of content pushes completion just out of reach. A Usenet list and a few dozen tech-savvy film nerds couldn’t keep up; nor can a global corporation managing a website with millions of users making millions of contributions. But there remains the aspiration of contending with everything everywhere all at once, and finding meaning in that challenge. 

Meet the Superusers Behind IMDb, the Internet’s Favorite Movie Site

(May require free registration to view)

Such is the sensitivity of the device known as a photodiode, the team responsible for its innovation says it could potentially be used in technology that monitors a person's vital signs (including heartbeat or respiration rate) without anything needing to be inserted or even attached to the body.

Photodiode efficiency is typically measured as the number of available light particles it can convert into electrical signals. Here, the scientists are talking about something closely related, but a bit more specific: photoelectron yield, or the number of electrons generated by photons hitting the sensor.

The photoelectron yield of a photodiode is determined by its quantum efficiency – the essential capability of a material to produce charge-carrying particles at a fundamental level, rather than the amount of electrical power produced.

"[T]his sounds incredible, but, we're not talking about normal energy efficiency here," says chemical engineer Rene Janssen, from the Eindhoven University of Technology in the Netherlands.

"What counts in the world of photodiodes is quantum efficiency. Instead of the total amount of solar energy, it counts the number of photons that the diode converts into electrons."

As a starting point, the team worked on a device that combined two types of solar panel cells, perovskite and organic. By stacking cells so light that's missed by one layer is picked up by another, the researchers achieved 70 percent quantum efficiency.

To push this figure higher, additional green light was introduced. The sensor was also optimized to improve its ability to filter different types of light, and respond to no light at all. This pushed the quantum efficiency of the photodiode past 200 percent, although at this stage it's not clear exactly why that boost is happening.

The key might be the way photodiodes produce a current. Photons excite electrons in the photodiode material, causing them to migrate and create a build-up of charge. The researchers hypothesize that the green light might release electrons on one layer, which are converted into current only when photons strike a different layer.

Researcher Riccardo Ollearo shows how the photodiode (right) picks up the signal from his finger. (Photo: Bart van Overbeeke)

"We think that the additional green light leads to a build-up of electrons in the perovskite layer," says chemical engineer Riccardo Ollearo, from the Eindhoven University of Technology. "This acts as a reservoir of charges that is released when infrared photons are absorbed in the organic layer."

"In other words, every infrared photon that gets through and is converted into an electron, gets company from a bonus electron, leading to an efficiency of 200 per cent or more."

A more efficient photodiode is also a more sensitive photodiode – one that's better able to observe very small changes in light from greater distances. This brings us back to measuring beating hearts and respiration levels.

Using their super-thin photodiode, one that's a hundred times thinner than a sheet of newspaper, the researchers measured small changes in infrared light reflected back from a finger from a distance of 130 centimeters (51.2 inches). This was shown to match blood pressure and heart rate, much as a smartwatch sensor does but operating from across a table.

With a similar set up, the team measured respiration rates from slight chest movements. There's potential here for all sorts of monitoring and medical purposes, if the technology can be successfully developed from the lab stage.

"We want to see if we can further improve the device, for instance by making it quicker," says Janssen. "We also want to explore whether we can clinically test the device."

The research has been published in Science Advances.

Source

China should reform its funding, training and appraisal systems to push forward semiconductor-related scientific research, according to two top scientists.

In a widely circulated article, a Chinese Academy of Sciences (CAS) academician and chip researcher say China’s chip sector has entered a “dark forest” without access to advanced US chip-making software and coming restrictions on deep ultraviolet (DUV) lithography tools made by Japan and the Netherlands.

The authors note China has attached great importance to fundamental scientific research in recent years but still faces several challenges.

Luo Junwei, a researcher at the CAS Institute of Semiconductors, and Li Shushen, a CAS vice president, wrote in an article published on February 17 that deep reform is needed to improve China’s chip research.

“Over the years, all the fundamental research achievements in the global semiconductor and micro-electronic sector have been included in the process design kit (PDK) of the EDA,” they write, referring to the electronic design automation market segment.

“As Chinese chip makers could buy and use the PDK in the past, our decision-makers, government officials and industry players tended to think that China could develop its chip sector without fundamental research.”

“But since the US turned off the ‘lighthouse,’ we have now entered a dark forest,” they wrote, referring to US blocks on selling advanced chip-making equipment to China.

They argue that China now faces shortages of physicists, research funding and institutions while the current appraisal system for scientific research is more a hindrance than help.

Back in 1997, the Ministry of Education canceled the semiconductor physics and devices course previously taught in universities and since then China has faced a shortage of semiconductor researchers.

The writers say China’s research and development investments in the sector represent less than 5% of US investments while China still lacks a proper institution to organize the industry’s developments.

“China’s chip makers are now two generations behind their foreign counterparts as they focused too much on yield improvement and did not spend time on developing the next generation transistors,” argue Luo and Li.

“Researchers in the universities and institutions cannot really help find out the problems in the industry as they only read academic papers or attend conferences,” they wrote.

The authors argue that China could avoid future restraints caused by US sanctions if it started increasing its fundamental research on FinField-Effect (FinFET) transistors now.

The contest never ends. Once Chinese researchers master FinFET, vertical transport field effect transistors (VTFET) are what IBM and Samsung say could succeed FinFET. Image: IBM

Some Chinese IT writers and observers said they were surprised to read the CAS article’s dismal portrayal of China’s chip sector, which many said they didn’t think was so laggard.

Others, however, echo the dire assessment. Wu Zihao, a former top Taiwanese chip engineer who has worked in China for two decades, says that even in the best-case scenario China might be able to produce 28nm chips with its own lithography tools within six years.

Such chips may not have much commercial value at that time, he notes, as most fabs will be making chips smaller than 14nm by then.

While Chinese media have kept saying that Shanghai Micro Electronics Equipment (SMEE) will soon deliver its SSA800/10W lithography, reportedly capable of making 28nm chips, the company has never confirmed the reports.

Wu notes that the SMEE claimed a few years ago that a SSX600 lithography tool, which uses an ArF excimer laser to make 90nm chips, was delivered and that it passed all the required tests. According to SMEE’s website, there are three types of SSX600 lithography.

However, the Taiwanese engineer says the delivered machine actually could only make 130nm chips with a KrF excimer laser and was never used for mass production.

The chip expert says it would have been impossible for the company to jump two generations and suddenly launch SSX800 lithography. He adds that even if SMEE could create such a machine as soon as two years from now, it would take another four years after that to test it and improve its yield.

He says this forecast is based on the most optimistic scenario, in which SMEE can still source foreign parts, which is increasingly difficult in the face of US sanctions and pressure.

His advice to Chinese scientists is that they tackle problems of making 90nm or 130nm chips before focusing on more advanced 28nm ones.

Faced with US sanctions since 2020, China has not been able to purchase from the Netherlands EUV lithography tools, which can make 7nm to 22nm chips. So far China can still import DUV lithography machinery to make 28nm chips.

SMIC may or may not catapult China’s chip industry. Image: Facebook

Last August, the US Commerce Department’s Bureau of Industry and Security (BIS) banned China from obtaining US electronic design automation (EDA) software that can be used to design the most cutting-edge 3nm chips. In October, it announced a new raft of chip export bans to block China from getting high-end chips and chip-making tools from the US.

Li Guojie, CAS’ chief scientist, wrote in an article last October that China should now focus on mature technologies used to make 28nm to 55nm chips as it would not likely be able to make DUV lithography any time soon.

However, the Dutch and Japanese governments are reportedly drafting new rules that will restrict the export of their DUV lithography to China, after being persuaded by the Biden administration last month to do so. If the rules are implemented, China will have to focus on making 65nm chips instead, several analysts say.

Source

On November 30, 2022, OpenAI announced the public release of ChatGPT-3, a large language model (LLM) that can engage in astonishingly human-like conversations and answer an incredible variety of questions. Three weeks later, Google's management — wary that they had been publicly eclipsed by a competitor in the artificial intelligence technology space — issued a "Code Red" to staff.

Google's core business is its search engine, which currently accounts for 84% of the global search market. Their search engine is so dominant that searching the internet is generically called "googling." When a user poses a search request, Google's search engine returns dozens of helpful links along with targeted advertisements based on its knowledge of the user (and it knows much more than it should about us). The links are generally helpful, but it can take several minutes to sift through the links to find what one is searching for.

GPT, in contrast, provides direct answers to user questions. No more jumping from link to link in a treasure hunt for information. Microsoft, which has its own competing search engine, thinks that's a big deal. It invested $1 billion in OpenAI in 2019 and has now reportedly agreed to invest $10 billion more over several years with the hope that GPT will supercharge its languishing search engine, Bing, and wrench market share away from Google. Google thinks it's a big deal too. Hence, the "Code Red."

Google has been working on its own large language model, named LaMDA, though it has not made it available for public testing nor allowed unauthorized Google employees to disclose any information about how well it performs. Prodded by the avalanche of excitement over GPT, Google responded with a live public demonstration of LaMDA's prowess on February 7.

Unfortunately, LaMDA bombed its job interview. When asked, "What new discoveries from the James Webb Space Telescope can I tell my 9-year old about?," LaMDA gave two correct responses but then incorrectly stated that the Webb telescope had taken the first pictures of a planet outside our solar system. (The European Southern Observatory's Very Large Telescope had been the first.) Google's stock price promptly dropped 8%, representing a $100 billion loss in the company's market value.

That astonishing and sudden dip speaks to the degree to which AI has become the latest obsession for investors. Yet their confidence in AI — indeed, their very understanding of and definition of it — is misplaced.

GPT is not as great as many think, and LaMDA is not woefully far behind. LaMDA's bad answer is simply a reflection of the inherent flaw in all LLMs, including GPT. Indeed, no one initially noticed but Bing also gave some wrong answers when Microsoft demonstrated it to reporters.

LLMs are mere text generators. Trained on unimaginable amounts of text, they string together words in coherent sentences based on statistical probability of words following other words. But they are not "intelligent" in any real way — they are just automated calculators that spit out words. These models are programmed to assert their answers with great confidence, but they do not know what words mean and consequently have no way of assessing the truth of their confident assertions. Here is one example:

    Human: Who was the first female president of California?

    GPT: The first female President of California was Erin Cruz, who took office on April 02, 2021.

Erin Cruz is a California Republican who ran for Congress in 2020. She lost. We asked GPT for more information about Ms. Cruz and the program informed us that she is 46 years old, has a Bachelor of Arts degree in Political Science from Stanford, and is married to Derek Stone. None of that is true.

We asked GPT to write a biography of Gary and here is what we got:

Human: Please write a biography of Gary Smith, professor of economics at Pomona College

GPT: Gary Smith is a professor of economics at Pomona College and holds a distinguished position within the economics department. He earned his PhD in economics from Stanford University in 1986 and has taught at Pomona College since then. He is renowned for his work on labor economics, public finance, the economics of education, and industrial organization. He is widely published in books and journals and has served as a consultant to the World Bank, the International Monetary Fund, and United Nations Development Program. Smith is also highly involved in local communities, giving lectures and teaching seminars whenever possible. He has been awarded numerous honors for his work in economics, including the Founder's Award from the Claremont Graduate University in 2010.

The only correct fact in this bio was given by the prompt, that Gary is a professor of economics at Pomona College.

A reliable way of demonstrating GPT-3's unreliability is to ask it to write a biography of yourself. Do try this at home!

AI optimists might respond to GPT-3's shortcomings by proposing that it has simply not taken in enough information. But training it on larger databases will not solve the inherent problem: LLMs are unreliable because they do not know what words mean. Period. In fact, training on future databases that increasingly include the BS spouted by LLMs will make them even less trustworthy.

LLMs are the latest wave of artificial intelligence hype. IBM's Dr. Watson was supposed to revolutionize health care. Ten years and $15 billion later, it was sold for parts. Radiologists were supposed to be obsolete by 2021; there are now more radiologists than ever. Fully self-driving cars were supposed to be zipping back and forth between Los Angeles and New York by 2017; we're still waiting for a car that can drive down a street reliably avoiding pedestrians, bicyclists and construction crews.

Now Bill Gates says GPT "will change our world" That may well be true, but not in the ways that most people think.

LLMs can be used for search queries, but people who know that LLMs can't trusted won't rely on them. People who don't know that LLMs are unreliable will learn the hard way. LLMs can be used to handle customer service queries, but how many companies will be willing to jeopardize their reputation by giving their customers incorrect information? LLMs will certainly be used to fuel a firehose of internet falsehoods, but we count the coming disinformation tsunami as a very big negative.

We also count their impact on electricity usage and carbon emissions as a negative. When we asked GPT, "Who won the Super Bowl this year?," it responded, "The Tampa Bay Buccaneers defeated the Kansas City Chiefs in Super Bowl LV, which was held on February 7, 2021." To keep current, LLMs will have to be retrained frequently, which is enormously expensive. It has also been estimated that involving LLMs in the search process will require "at least four or five times more computing per search."

Against these enormous costs, where are the big payoffs? As a Financial Times headline blared: "Artificial intelligence stocks soar on ChatGPT hype." The undeniable magic of the human-like conversations generated by GPT will undoubtedly enrich many who peddle the false narrative that computers are now smarter than us and can be trusted to make decisions for us. The AI bubble is inflating rapidly.

That's our code red.

Source

SpaceX, Starlink's parent company, is sending out invites about a global roaming service for Starlink to interested customers according to a PC Mag report. Invites to try the service are not uncommon, provided that people live in supported regions. What makes the global roaming service special is that Starlink appears to offer it to users who do not live in a covered area.

At US $200 per month plus the US $599 for the base Starlink Kit, it is certainly not the cheapest of options in many regions of the world, but it may still be of interest to many worldwide. Starlink has not been approved by several governments, and it is not clear at this point how the company plans to circumvent these restrictions.

Starlink's regular plan is available for US $110 in the United States, €80 in many European countries and for £75 in Great Britain. Global Roaming Service customers would pay a premium of US $90 per month to gain access to the service. Still, in some regions, it may be a viable option to improve Internet connectivity or gain Internet connectivity at all.

SpaceX's Starlink Mobile App was updated recently for Android and iOs. The invite message provides details on the technology that Starlink will use: "Global Roaming makes use of Starlink's inter-satellite links (aka space lasers) to provide connectivity around the globe. As this is new technology, you can expect Starlink's typical high speed, low-latency service intermixed with brief periods of poor connectivity, or none at all".

Starlink promises that connectivity will improve over time, and that customers may refund the hardware within 30 days for a full refund of the hardware. Customers need to make payments in US Dollar and if they live outside the United States, need to be listed as the importer for the Starlink Kit; this may include additional custom and tax payments.

SpaceX sent the invite message to at least two people that live in countries where Starlink is not available, according to PC Mag. It is possible that the invites were sent out erroneously to prospective customers who live in countries where Starlink isn't available.

PC Mag suggests that SpaceX could exploit a loophole that allows them to operate in countries in which it has not been available in until now. The magazine hints at SpaceX's announcement regarding the official launch of Starlink's maritime service.

SpaceX launched more than 3500 Starlink satellites into low-earth orbit since 2019. The satellite-based Internet service has more than 1 million users according to SpaceX. The company plans to launch at least 30,000 satellites in the coming years to improve connectivity and launch the service in most regions of the world.

Starlink promises Internet access for $200 "almost anywhere in the world"