Trending February 2024 # Nvidia Tries To Rein In Chatgpt’s Overactive Imagination # Suggested March 2024 # Top 5 Popular

You are reading the article Nvidia Tries To Rein In Chatgpt’s Overactive Imagination updated in February 2024 on the website We hope that the information we have shared is helpful to you. If you find the content interesting and meaningful, please share it with your friends and continue to follow and support us for the latest updates. Suggested March 2024 Nvidia Tries To Rein In Chatgpt’s Overactive Imagination

Nvidia, the GPU mega weight that helps train language models like ChatGPT, has recently launched ‘NeMo Guardrails’ — an open-source software designed to keep AI chatbots on the straight and narrow.

According to the company, the software aims to keep responses on topic, improve data security, and combat random spurts of inaccurate information commonly known as AI ‘hallucinations’.

While this doesn’t satisfy the six-month AI development pause that tech leaders like Elon Musk and Steve Wozniak are calling for, it does aim to address some major issues the technology faces today.

Nvidia Releases ‘NeMe Guardrails’ to Tackle AI Hallucination

AI tools like ChatGPT, Google Bard, and Bing Chat are capable of responding to just about any prompt fired at them. But this doesn’t mean their responses should always be trusted.

When put to the test, OpenAI’s ChatGPT is consistently found to give inaccurate answers, with the chatbot routinely failing at basic math, going off script, and spouting out content that seems straight-up implausible.

Nvidia — the supercomputing giant that’s responsible for training AI tools like ChatGPT — is aware of its tendency to hallucinate and has created NeMe Guardrails in an attempt to improve the accuracy and safety of the technology.

“Safety in generative AI is an industry-wide concern. NVIDIA designed NeMo Guardrails to work with all LLMs, such as OpenAI’s ChatGPT.” – Nvida blog post

NeMe helps developers make sure language models stick to their requirements by helping them to instate topical, safety, and security guardrails.

In Nvidia’s own words, the software’s topical rails aim to ‘prevent apps from veering off into undesired areas’, while its security guardians ‘ensure apps respond with accurate, appropriate information’.

Finally, its security guardrails work by preventing the tools from connecting to unsafe third-party apps that may be culpable of compromising private information.

But how does the software limit chatbot delusion? According to Nvidia, the software uses a second logic learning model (LLM) to fact-check the answers of the first one. If the second LLM doesn’t come up with matching answers, the response is deemed to be a hallucination before it’s sent to the user.

Who Can Use NeMe Guardrails?

Since NeMo Guardrails runs on open-source technology, its able to be used by any enterprise app developer looking to add extra safeguards to its chatbot.

Programmers are able to use the language to create custom rules for their AI model, implementing as many guardrails as they see fit.

The software is being incorporated into the NVIDIA NeMo  framework which includes everything you would need to train and tune a language model and is currently available on GitHub.

Are These Guardrails Enough to Keep Users Safe?

Nvidia’s new software represents an important development in chatbot accuracy.

However, while NeMe Guardrails was designed to keep AI-generated content on track while protecting users from security risks, it fails to address instances of “bias” and “deception” cited in a recent complaint to the Federal Trade Commission (FTC) by the Center of AI and Digital Policy (CAIDP).

“We look forward to the good that will come from making AI a dependable and trusted part of the future.” – Nvidia blog post

After pivoting its focus to AI technology, Nvidia has profited heavily from the explosion of tools like ChatGPT and Microsoft’s Bing Chat, meaning it’s unlikely to heed the calls of concerned voices such as Elon Musk and Steve Wozniak to slow down.

Ultimately, while some AI skeptics may fear NeMe Guardrails don’t go far enough, the software does give developers a solid framework to follow. What’s more, with the US rolling out AI controls a lot slower than its European counterparts, we think that any attempt to improve and regulate chatbot technology represents a promising step in the right direction.

You're reading Nvidia Tries To Rein In Chatgpt’s Overactive Imagination

Chatgpt’s Accuracy Has Gotten Worse, Study Shows

A pair of new studies presents a problematic dichotomy for OpenAI’s ChatGPT large language model programs. Although its popular generative text responses are now all-but-indistinguishable from human answers according to multiple studies and sources, GPT appears to be getting less accurate over time. Perhaps more distressingly, no one has a good explanation for the troubling deterioration.

A team from Stanford and UC Berkeley noted in a research study published on Tuesday that ChatGPT’s behavior has noticeably changed over time—and not for the better. What’s more, researchers are somewhat at a loss for exactly why this deterioration in response quality is happening.

To examine the consistency of ChatGPT’s underlying GPT-3.5 and -4 programs, the team tested the AI’s tendency to “drift,” i.e. offer answers with varying levels of quality and accuracy, as well as its ability to properly follow given commands.  Researchers asked both ChatGPT-3.5 and -4 to solve math problems, answer sensitive and dangerous questions, visually reason from prompts, and generate code.

[Related: Big Tech’s latest AI doomsday warning might be more of the same hype.]

In their review, the team found that “Overall… the behavior of the ‘same’ LLM service can change substantially in a relatively short amount of time, highlighting the need for continuous monitoring of LLM quality.” For example, GPT-4 in March 2023 identified prime numbers with a nearly 98 percent accuracy rate. By June, however, GPT-4’s accuracy reportedly cratered to less than 3 percent for the same task. Meanwhile, GPT-3.5 in June 2023 improved on prime number identification in comparison to its March 2023 version. When it came to computer code generation, both editions’ ability to generate computer code got worse between March and June.

These discrepancies could have real world effects—and soon. Earlier this month, a paper published in the journal JMIR Medical Education by a team of researchers from NYU indicates ChatGPT’s responses to healthcare-related queries are ostensibly indistinguishable from human medical professionals when it comes to tone and phrasing. The researchers presented 392 people with 10 patient questions and responses, half of which came from a human healthcare provider, and half from OpenAI’s large language model (LLM). Participants had “limited ability” to distinguish human- and chatbot-penned responses. This comes alongside increasing concerns regarding AI’s ability to handle medical data privacy, alongside its propensity to “hallucinate” inaccurate information.. 

Academics aren’t alone in noticing ChatGPT’s diminishing returns. As Business Insider notes on Wednesday, OpenAI’s developer forum has hosted an ongoing debate about the LLM’s progress—or lack thereof. “Has there been any official addressing of this issue? As a paying customer it went from being a great assistant sous chef to dishwasher. Would love to get an official response,” one user wrote earlier this month.

[Related: There’s a glaring issue with the AI moratorium letter.]

OpenAI’s LLM research and development is notoriously walled off to outside review, a strategy that has prompted intense pushback and criticism from industry experts and users. “It’s really hard to tell why this is happening,” tweeted Matei Zaharia, one of the ChatGPT quality review paper’s co-authors, on Wednesday. Zaharia, an associate professor of computer science at UC Berkeley and CTO for Databricks, continued by surmising that reinforcement learning from human feedback (RLHF) could be “hitting a wall” alongside fine-tuning, but also conceded it could simply be bugs in the system.

So, while ChatGPT may pass rudimentary Turing Test benchmarks, its uneven quality still poses major challenges and concerns for the public—all while little stands in the way of their continued proliferation and integration into daily life.

Lindows Tries To Take The Geek Out Of Linux

Computer users have grown spoiled over the years.

Earthlink founder Sky Dayton likes to tell the story of how it took him 80 hours to log onto the Internet the first time back in 1993. Now, when someone buys a computer, the modem and connection software come preloaded. Just plug in the cable, and it will locate the connection and configure itself.

Other features have gotten easier as well. In fact Windows has gotten so good with its plug-n-play that few users today even know how to edit the registry or resolve a basic conflict.

While Windows is good for untrained users, for those who want to get in under the hood and tinker with the source code, there is Linux.

Users can easily spend hours working out just how to get a device driver to work properly. While that might be fun for some, it also limits Linux market penetration. So, although Linux is giving Windows a run in the low- and mid-range server market, it has one thirtieth the number of users as Windows on the desktop.

Michael Robertson, founder of chúng tôi is tackling that problem with his new company, Lindows, Inc. Other Linux vendors have strong server offerings. Lindows goes straight for the users, offering a desktop version of Linux that installs in 10 minutes and also makes it easy for the untrained to download and install the applications they need.

Lindows is a privately held company based in San Diego. It develops and markets LindowsOS, a version of Debian Linux. Unlike Red Hat or SuSE, which offer users a choice of GUIs, Lindows comes with just the KDE desktop. If the user is sophisticated enough to know the difference between the KDE and Gnome desktops, and have a preference for Gnome, LindowsOS is not for them. But for someone who wants a low-cost, easy-to-use alternative to Windows or MacIntosh OSX, LindowsOS can get them up and running much faster.

The other way to get LindowsOS is pre-installed on a desktop, laptop or hard drive. Lindows sells webstations. Walmart even has six Microtel PC running Lindows. PC Club has Lindows laptops. And Seagate offers the OS preloaded on some drives at no additional cost.

Getting Down to Business

Most users, however, dont care what operating system they use. What they want is applications.

”Plenty of applications run on Linux,” says Mike Silver, a vice president at industry analyst giant Gartner, Inc., based in Stamford, Ct. ”The question is do they have the one I need with the functions I need.”

Here, again, Lindows makes it easier.

Now, a developer knows that there are more than 77,000 open source projects hosted on the chúng tôi site, plus plenty of other free software over at the KOffice and KDE sites. But most users don’t want to prowl around these places trying to decipher what they might need.

Lindows approach is to ship its OS with just the most essential software that most users will need. This includes a browser/email (Mozilla), an office suite (OpenOffice), instant messaging (Gaim) and some games and multimedia players.

To further simplify matters, the software is called by its generic name. For example the desktop icon for browser says ”browser” not ”Mozilla 1.6”. The CD Player is entitled just that, and so on.

The preloaded applications are enough for most people to get started, but there is plenty more software available.

Continue on to the next page to find out who Lindows is right for…

Lenovo Tries To Reinvent The Laptop (Again) With The Thinkpad X1 Fold

The branding may not have changed, but the tech certainly has. This updated X1 Fold is clearly more polished and streamlined than the version I saw at CES 2023, with a larger screen, new use cases, and enhanced tech. But is it ready to go mainstream?

Laptop, tablet, and more

The X1 Fold’s 16.3in display can be used at full-size either as a tablet or in landscape or portrait orientation thanks to its matching collapsible kickstand.

Dominic Preston / Foundry

But it also folds in half, giving you a 12in laptop with two screens – usable either in book form as a makeshift ereader, in laptop form with an on-screen keyboard, or using the optional keyboard (complete with Lenovo’s trademark TrackPoint and a haptic touchpad), which can itself be used either detached via Bluetooth, or magnetically attached to the lower screen to create a true traditional laptop setup.

The typing experience is better than you’d expect from a keyboard only held on with magnets. It doesn’t feel loose or flexible, with a solidity to the magnetic grasp that makes typing surprisingly satisfying.

That means the ThinkPad X1 Fold can serve as laptop, tablet, and all-in-one desktop PC, making it among the most versatile Windows 11 devices around. That’s if you have the compatible keyboard and case of course, which come as standard in Europe but are an optional extra if you’re in the US.

It’s portable too: the tablet itself weighs just 1.3kg/2.8lbs, rising to 1.9kg/4.2lbs if you include the stand and keyboard – which stack together when folding, for an overall package that has the same footprint as a 12in laptop, albeit a fair bit thicker. It’s also fiddlier and slower to pack up than simply closing a laptop lid,  but ultimately that feels like a small price.

Dominic Preston / Foundry

Lenovo has clearly thought carefully about making the most of each orientation. The X1 Fold is equipped with a surplus of microphones, speakers, and USB-C/Thunderbolt 4 ports, so that in any configuration you’ll get stereo sound, dual microphones, and at least two accessible USB-C ports.

It’s not short on power either. 12th-gen Intel U-series chips power the laptop/tablet hybrid, with a choice of i5 or i7 silicon. Up to 32GB RAM and 1TB storage add to the sense that, when properly kitted out, this could be a powerful productivity device – though with no discrete GPU option it won’t suit creative power users.

Lenovo has opted for an OLED panel, and promises 100% DCI-P3 colour coverage and support for Dolby Vision HDR content. It supports Lenovo’s Wacom-powered stylus too, while a thin bezel keeps it looking sharp.

A new bell-shaped hinge has been introduced along with the larger display, and it allows the two sides of the device to remain flush when closed, something Samsung’s phones still can’t. The crease is subtler too, and hard to spot unless you know where to look .

Fold vs Fold

After a slow start, folding phones are beginning to draw mainstream appeal, led by strong sales for last year’s Samsung Galaxy Z Flip 3 and Fold 3, both bolstered this August by updated fourth-gen models.

But it took Samsung three generations to begin to generate interest beyond enthusiasts, and Lenovo is still only on attempt number two. This is a clear progression from the prototype-y first-gen X1 Fold, but is the market really ready for a laptop screen that folds?

Dominic Preston / Foundry

That’s not even touching on the question of durability. Many buyers upgrade their phone every two years, if not sooner, but laptops are expected to last a little longer.

Lenovo says the device as a whole meets the military-grade MIL-STD-810H standards for toughness, and that the hinge itself can survive 30,000 folds – enough for five years of use if you fold and unfold 16 times a day, though depending on your usage you might fold it much more often than that.

Samsung’s success so far is also in large part down to price cuts last year, and it’s the $2,499/€2,999 start price that will likely be the biggest obstacle to success for the X1 Fold when it goes on sale this November.

That’s with the lowest specs available, and in the US it doesn’t include either the keyboard or stand. Lenovo hasn’t yet said how much those will add to the cost, but expect the full kit to be prohibitively expensive.

Dominic Preston / Foundry

It might help that Lenovo is no longer the only player in the space. Asus has joined them with its ZenBook 17 Fold OLED – first revealed in January at CES, but just yesterday confirmed to launch in December for a whopping $3,999. Suddenly the ThinkPad X1 Fold looks positively affordable.

Buyers still seem nervous about paying double for a phone that folds, so spending two or even three times as much as a typical laptop may well rankle – especially with doubts about durability.

If you’re still a skeptic, Lenovo has also launched a few more traditional products this week at IFA, including the affordable P11 and P11 Pro Android tablets, and the first-ever Chromebook with a 120Hz display.

Also find out what won our Best of IFA 2023 awards.

Refik Anadol On How Ai ‘Imagination’ Elevates Memory With Nfts

On June 25, 1949, the British neurologist Geoffrey Jefferson gave a lecture to the Royal College of Surgeons of England entitled The Mind of Mechanical Man. It may be surprising that machine intelligence was the subject of much debate in Jefferson’s time, with some describing the 1940s as the period in which artificial intelligence was born following the development of cybernetics. Jefferson’s ideas about the intersection of human and machine were ahead of their time and even impressed the great Alan Turing with their prescience and clarity.

“[N]ot until a machine can write a sonnet or a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain — that is, not only write it but know that it had written it,” Jefferson said in his lecture. “No mechanism could feel (and not merely artificially signal, an easy contrivance) pleasure at its successes, grief when its valves fuse, be warmed by flattery, be made miserable by its mistakes, be charmed by sex, be angry or miserable when it cannot get what it wants.”

The question of artificial intelligence

Whether they know it or not, critics of artificial intelligence’s application in the art world — and by extension, the world of NFTs — employ a version of Jefferson’s argument when they decry that the technology takes something away from the creative “soul” of artists and their work.

Of course, this could be assuming the consequent: Simply increasing the computational and adaptive powers of computing systems and calling it AI might not necessarily mean that these systems experience the drive to curiosity and sense of internal self-awareness that humans experience. Then again, the internal experience of consciousness and ego could also be an illusion. But to modern AI developers, thinking of humans as a special case might be outdated.

AI as an art tool

“In every chapter of humanity, endeavors and innovations and discoveries always light up some dinosaur views,” retorted Refik Anadol, the pioneering visual artist whose work sits at the intersection of digital media and machine learning, in an interview with nft now. “And I think that’s very normal. But to me, there’s an artist; there’s a desire. There’s a prompt; there’s a request; there is an input. I think this is pure collaboration — imagination with a machine.”

Anadol is well-known in the NFT space and beyond for his immersive (and often interactive) AI-infused pieces, including Melting Memories, a project inspired by his uncle’s Alzheimer’s diagnosis, An Important Memory for Humanity, which dropped in April of this year, and Unsupervised — Machine Hallucinations, which trained an AI model on the metadata of the Museum of Modern Art’s collection.

Three images from Unsupervised, via Refik Anadol

The idea of embedding media arts into architecture through data and computation has long driven Anadol’s approach to his work. But his love for computers came even earlier when his parents brought home a Commodore 64 for his eighth birthday, which enabled him to dive into the world of video games. Later that year, Anadol saw Blade Runner for the first time, which changed how he would imagine the future of humanity.

“As a child, I didn’t strive to see the optimism in life,” Anadol said. “I found much more optimism in [that] movie. I think that’s how I started to connect imagination and computation in my mind. Because to me, games were where we could escape or find new meaning, like creating a new world around a machine’s mind.”

Anadol went on to study Visual Communication Design at Istanbul Bilgi University, where he encountered a professor from Aalto University in Finland who taught Anadol how to start doing visual programming with a software called Pure Data. The programming language allows for interaction with algorithms and proximity sensors like microphones. 

“Data is still the pigment — But now, the brush can think.”

Refik anadol

“That’s when I saw, for the very first time, the potential of programming invisible signals of data,” Anadol recalled. “It was a beautiful signal transforming into black and white points and dots and lines. And that’s how I got this idea of [thinking of] data as a pigment, and that algorithms can become our brush.”

Before Anadol left Turkey to go to the States, he created a data sculpture media wall in Istanbul in collaboration with the architect Alper Derinboğaz, who had recently returned to Turkey after finishing a graduate degree at UCLA. Derinboğaz’s mentor, the famous Greg Lynn, inspired Anadol with Animate Form, a book in which Lynn argues that the future of architecture is not static.

“I was inspired by data as a material,” Anadol explained.

Anadol then went to study at UCLA, choosing the university because it housed professors like Casey Reas. Reas is well-known for creating Processing, an open-source creative coding community that enabled Anadol to learn Javascript for the first time.

“He’s a true hero and true pioneer in the field,” Anadol said. “He opened my mind.”

Data as an expression of humanity

Anadol believes that data is a form of memory and that this memory can have many different shapes, colors, and textures. The main objective Anadol has in his artistic practice is to make art for anyone of any age or background.

“This idea of finding the language of humanity is a very challenging practice,” Anadol elaborated. “Archives are the memories of humanity. They have this cultural and historical context. That’s why I focus on urban spaces in my work, specifically looking for patterns of memories that belong to everyone. I look at these memories of humanity and how they can merge with the powerful creativity of AI.”

When discussing common fears that AI might replace the vital force of human creativity, Anadol casually dismissed them, emphasizing the collaborative nature of the technology.

Machine Hallucinations, via Refik Anadol

“I’ve found that [AI] is an extension of my consciousness. It’s an extension of my imagination,” he said, “I see AI becoming this extension of the human mind. When I did Wind of Boston, I was [telling people that] the data is here becoming pigments and the algorithm can become a kind of a brush. Data is still the pigment — but now, the brush can think.”

Anadol generally agrees with artists like Claire Silver, who believe that because AI enables those who aren’t talented in drawing or painting to create art, skill and talent will slowly take a backseat to concept and vision. And while he doesn’t find programs like DALL-E 2 and MidJourney to be particularly groundbreaking in the nature of their algorithms, the egalitarian approach to creating AI art that these systems provide inspires him greatly.

Anadol’s coming projects are unsurprisingly ambitious. He’s released some tantalizing images on his Twitter account lately that hint at what fans of his art can expect, and told nft now that this next release will be an iteration of his DATALAND series.

New “beings” discovered at DATALAND! Can’t wait to unfold this unique narrative with you all soon! chúng tôi Refik Anadol (@refikanadol) August 24, 2023

“Basically, we are also exploring Stable Diffusion in a very custom way,” Anadol teased. “Six months ago, I was using DALL-E 2, and it was a massive trigger for my imagination. And for six months, I’ve been prompt engineering many concepts. DATALAND is completely designed by machine-assisted imagination. It will be a breakthrough in our journey as a studio, but also for the Web3 community.”

Anadol plans to continue using AI to create art while expanding its use cases in the context of Web3. And it’s having a galvanizing effect. Needless to say, his passion for the future of the space — and all of the tools at his disposal to potentially revolutionize it — is contagious.

“As a person inspired by innovation and discovery every single moment of my life, I can say that this is one of the most ambitious and truly culturally Web3 projects,” Anadol said. “It is not something we are used to [seeing] in the space. It’s more than an NFT collection — it’s an experience in life first.”

It’s Not Your Imagination: 2023 Has Been Abnormally Hot So Far

A famous 1936 image shows the people of Nebraska sprawled on the capitol lawn to escape record-breaking temps that had turned their AC-free homes into slow roasters. These vintage photos can make modern summer heat seem cool in comparison. But it isn’t. Average temperatures today are warmer than they were a century ago. And this week, the National Oceanic Atmospheric Administration (NOAA) declared the first six months of 2023 the second warmest on record. First place went to 2024.

Global land and ocean temperature anomolies NOAA

What, you say, no records broken this year? Not so fast. Researchers actually expected 2023 to be cooler than 2024. Because this year, unlike last, is not an El Niño year. El Niño, otherwise known as the southern oscillation, causes circulation patterns that tend to lead to warmer global temperatures. Absent this cyclical climate phenomenon, temperatures tend to be cooler. But as NOAA’s data and maps illustrate, that hasn’t made much of a difference this year—it’s still hot as heck out there.

June temperature across global land and ocean surfaces was 1.48°F above the 20th century average of 59.9°F. NOAA

The year’s warmth is just another piece of evidence supporting what climate scientists have said for over a decade. Put simply, the climate has lost its chill. Due to human behaviors—namely, releasing greenhouse gasses such as carbon dioxide and methane into the atmosphere—the climate is warming up, and that’s leading to increasingly volatile weather.

If climate change simply meant hotter temperatures, that wouldn’t be great—but it would be something we could probably adapt to. But that isn’t what climate change means. It means what some researchers have nicknamed “global weirding.” It means increasing rainfall in the northeast, patterns which we already saw evidenced in Canadian floods earlier this year.

“You know, I’ve often said in Canada that our greatest attribute, in terms of the weather, is that the weather hits and runs,” David Phillips, the senior climatologist with Environment and Climate Change Canada (the agency tasked with implementing the Canadian government’s environmental agenda) told PopSci at the height of the flooding earlier this year. But now bad weather is lingering—dropping buckets when it used to only leave droplets, increasing instances of flooding. That’s all in line with predictions about how human caused climate change will continue to play out.

Global weirding can also mean more droughts—something we’re currently seeing in Montana and the Dakotas. Warmer waters also mean more coral reefs are dying: The Great Barrier reef recently bleached for the second time in just two years.

“The thing you have to bear in mind is, prior to human-induced warming of the Earth’s climate, bleaching was only ever a very localized and short-term event,” Sean Connolly, an associate professor at Australia’s James Cook University and a part of the ARC Center of Excellence Coral for Reef Studies located at the university, told PopSci.

Selected significant climate anomalies, June 2023 NOAA

Canada’s British Columbia is in a state of emergency as more than 160 wildfires burn across the province. Because of the nature of climate forecasting, it’s difficult to say that climate change is the culprit—but Lori Daniels, an associate professor of forest ecology at the University of British Columbia, told the Canadian publication the Globe and Mail:

North of Canada, the arctic sea ice is continuing to decline. Sea ice grows in the winter, reaching its greatest coverage in late May/early June. This year, the amount of sea ice was the sixth least it’s ever been since records began in 1979.

Update the detailed information about Nvidia Tries To Rein In Chatgpt’s Overactive Imagination on the website. We hope the article's content will meet your needs, and we will regularly update the information to provide you with the fastest and most accurate information. Have a great day!