Daily Archives: March 14, 2024

The race to reach artificial general intelligence is getting intense among the tech giants, but its arrival will not happen as a moment in time, according to Amazons vice president of AGI.

Its very unlikely that theres going to be a moment in time when you suddenly decide, oh AGI wasnt (here yesterday) but its here today, said Vishal Sharma during a fireside chat at SXSW 2024 in Austin, Texas. Thats probably not going to happen.

Instead, he sees it as a journey of continuous advances. His comments echo Google DeepMinds six levels of AGI, where models go up one level as they progressively exhibit more AGI characteristics.

Meanwhile, there are hurdles to overcome. For one, people still do not agree on a precise definition of AGI. If you ask 10 experts about AGI, you will get 10 different explanations," he said.

Another is the ethical challenges models face. For Sharma, they fall in three buckets: Veracity since the models can hallucinate or make things up safety (intense red-teaming is needed), and controllability, in which inputting broadly similar prompts or queries can result in broadly similar outcomes.

A popular technique to mitigate hallucinations is Retrieval-Augmented Generation (RAG) in which the model is given, or provided access to, additional content or data from which to draw its answers. Sharma said RAG is still the best technique to fight hallucinations today.

Related:DeepMind Co-founder on AGI and the AI Race - SXSW 2024

However, he mentioned that there is another school of thought that believes its just a matter of time until the models become capable enough where these truths will be woven into the model themselves.

As for his views on open vs. closed models, Sharm said one of Amazons leadership principles is that success and scale bring broad responsibility and this applies to both types of models.

He emphasized the need to be flexible since generative AI remains fairly new and unforeseen opportunities and challenges could arise. Sharma said that when the internet began maturing, it brought new challenges that people did not think of before, such as cyber bullying.

We have to be adaptable, Sharma said.

He also thinks that just as the rise of semiconductors ushered in Moores Law and the network of networks led to Metcalfes Law, generative AI could lead to a new principle as well.

Credit: Amazon

He sees a time when AI will be broadly embedded into daily life as a helpful assistant, while staying in the background.

Sharma said Alexas Hunches are already one sign of this future. With Hunches, Alexa learns your routine say locking the back door at 9 p.m. every night and if you fail to do that one night, it will send an alert.

Related:EU AI Act Would Scrutinize Many General AI Models SXSW 2024

He said Amazons Astro is an example of an embodied AI assistant. The $1,600 household robot is used for home monitoring. You can ask it to check on people or specific rooms in the house. It alerts you if it sees someone it does not recognize or hears certain sounds. Astro can also throw treats to your dog through an accessory that is sold separately.

To be sure, todays models still have room for improvement whether in performance or economics. But Sharma believes advancements will lead to an age of abundance through the fusion of use cases that will become possible.

You should bet on AI, he said. You should not bet against it.

View post:

Amazon's VP of AGI: Arrival of AGI Not 'Moment in Time' SXSW 2024 - AI Business

Corvids are a family of birds that are known to be astonishingly accomplished at showing self-awareness and problem-solving via the use of tools. Such traits are generally considered to be extremely rare in the animal kingdom, as there's only ourselves and a handful of other species that can do all of this. However, you'd never think for one moment that any corvid is a human: We recognise the fact they are smart but not truly intelligent, or certainly not to the extent that we are.

And it's the same when it comes to artificial intelligence, the biggest topic in the world of computing and tech right now. While we've seen incredibly rapid progress in certain areas, such as generative AI video, nothing produced by the likes of ChatGPT, Stable Diffusion, or Copilot gives us the impression that it's true, human-like intelligence. Typically classed as weak or narrow AI, such systems aren't self-aware nor are they problem-solving, as such; they're basically enormous probability calculators, heavily reliant on the datasets used to train them.

Pinning down exactly what is meant by the phrase human intelligence is something that the scientific community has battled over for centuries, but in general, we can say it's the ability to recognise information or infer it from various sources, and then use it to plan, create, or problem solve through logical reasoning or abstract thinking. We humans do all of this extremely well, and we can apply it in situations that we've not had experience or prior knowledge of.

Getting a computer to exhibit the same capabilities is the ultimate goal of researchers in the field of artificial general intelligence (AGI): Creating a system that is able to conduct cognitive tasks just as well as any human can, and hopefully, even better.

What is artificial general intelligence?

This is a computer system that can plan, organise, create, reason, and problem-solve just like a human can.

The scale of such a challenge is rather hard to comprehend because an AGI needs to be able to do more than simply crunch through numbers. Human intelligence relies on language, culture, emotions, and physical senses to understand problems, break them down, and produce solutions. The human mind is also fragile and manipulable and can make all kinds of mistakes when under stress.

Sometimes, though, such situations generate remarkable achievements. How many of us have pulled off great feats of intelligence during examinations, despite them being potentially stressful experiences? You may be thinking at this point that all of this is impossible to achieve and surely nobody can program a system to apply an understanding of culture, utilise sight or sound, or recall a traumatic event to solve a problem.

It's a challenge that's being taken up by business and academic institutions around the world, with OpenAI, Google DeepMind, Blue Brain Project, and the recently completed Human Brain Project being the most famous examples of work conducted in the field of AGI. And, of course, there's all the research being carried out in the technologies that will either support or ultimately form part of an AGI system: Deep learning, generative AI, neural language processing, computer vision and sound, and even robotics.

As to the potential benefits that AGI could offer, that's rather obvious. Medicine and education could both be improved, increasing the speed and accuracy of any diagnosis, and determining the best learning package for a given student. An AGI could make decisions in complex, multi-faceted situations, as found in economics and politics, that are rational and beneficial to all. It seems a little facile to shoehorn games into such a topic but imagine a future where you're battling against AGI systems that react and play just like a real person but with all of the positives (comradery, laughter, sportsmanship) and none of the negatives.

Not everyone is convinced that AGI is even possible. Philosopher John Searle wrote a paper many decades ago arguing that artificial intelligence can be of two forms, Strong AI and Weak AI, where the difference between them is that the former could be said to be consciousness whereas the latter only seems like it does. To the end user, there would be no visible difference, but the underlying system certainly isn't the same.

The way that AGI is currently progressing, in terms of research, puts it somewhere between the two, though it's more weak rather than strong. Although this may seem like it's just semantics, one could take the stance that if the computer only appears to have human-like intelligence, it can't be considered to be truly intelligent, ultimately lacking what we consider to be a mind.

AI critic Hubert Dreyfus argues that computers are only able to process information that's stored symbolically and human unconscious knowledge (things that we know about but never directly think about) can't be symbolically stored, thus a true AGI can never exist.

A fully-fledged AGI is not without risks, either. At the very least, the widespread application of them in specific sectors would result in significant unemployment. We have already seen cases where both large and small businesses have replaced human customer support roles with generative AI systems. Computers that can do the same tasks as a human mind could potentially replace managers, politicians, triage nurses, teachers, designers, musicians, authors, and so on.

Perhaps the biggest concern over AGI is how safe it would be. Current research in the field is split on the topic of safety, with some projects openly dismissive of it. One could argue that a truly artificial human mind, one that's highly intelligent, may see many of the problems that humanity faces as being trivial, in comparison to answering questions on existence and the universe itself.

Building an AGI for the benefit of humanity isn't the goal of every project at the moment.

Despite the incredible advances in the fields of deep learning and generative AI in recent years, we're still a long way off from having a system that computer scientists and philosophers universally agree on having artificial general intelligence. Current AI models are restricted to very narrow domains, and cannot automatically apply what they have learned into other areas.

Generative AI tools cannot express themselves freely through art, music, and writing: They simply produce an output from a given input, based on probability maps created through trained association.

Whether the outcome turns out to be SkyNet or HAL9000, Jarvis or Tars, AGIs are still far from being a reality, and may never do so in our lifetimes. That may well be a huge relief to many people, but it's also a source of frustration for countless others, and the race is well and truly on to make it happen. If you've been impressed or dismayed by the current level of generative AI, you've seen nothing yet.

Read the original here:

What is general intelligence in the world of AI and computers? The race for the artificial mind explained - PC Gamer

Join leaders in Boston on March 27 for an exclusive night of networking, insights, and conversation. Request an invite here.

Last week, Anthropic unveiled the 3.0 version of their Claude family of chatbots. This model follows Claude 2.0 (released only eight months ago), showing how fast this industry is evolving.

With this latest release, Anthropic sets a new standard in AI, promising enhanced capabilities and safety that for now at least redefines the competitive landscape dominated by GPT-4. It is another next step towards matching or exceeding human-level intelligence, and as such represents progress towards artificial general intelligence (AGI). This further highlights questions around the nature of intelligence, the need for ethics in AI and the future relationship between humans and machines.

Instead of a grand event, Anthropic launched 3.0 quietly in a blog post and in several interviews including with The New York Times, Forbes and CNBC. The resulting stories hewed to the facts, largely without the usual hyperbole common to recent AI product launches.

The launch was not entirely free of bold statements, however. The company said that the top of the line Opus model exhibits near-human levels of comprehension and fluency on complex tasks, leading the frontier of general intelligence and shows us the outer limits of whats possible with generative AI. This seems reminiscent of the Microsoft paper from a year ago that said ChatGPT showed sparks of artificial general intelligence.

The AI Impact Tour Boston

Were excited for the next stop on the AI Impact Tour in Boston on March 27th. This exclusive, invite-only event, in partnership with Microsoft, will feature discussions on best practices for data integrity in 2024 and beyond. Space is limited, so request an invite today.

Like competitive offerings, Claude 3 is multimodal, meaning that it can respond to text queries and to images, for instance analyzing a photo or chart. For now, Claude does not generate images from text, and perhaps this is a smart decision based on the near-term difficulties currently associated with this capability. Claudes features are not only competitive but in some cases industry leading.

There are three versions of Claude 3, ranging from the entry-level Haiku to the near expert level Sonnet and the flagship Opus. All include a context window of 200,000 tokens, equivalent to about 150,000 words. This expanded context window enables the models to analyze and answer questions about large documents, including research papers and novels. Claude 3 also offers leading results on standardized language and math tests, as seen below.

Whatever doubt might have existed about the ability of Anthropic to compete with the market leaders has been put to rest with this launch, at least for now.

Claude 3 could be a significant milestone towards AGI due to its purported near-human levels of comprehension and reasoning abilities. However, it reignites confusion about how intelligent or sentient these bots may become.

When testing Opus, Anthropic researchers had the model read a long document in which they inserted a random line about pizza toppings. They then evaluated Claudes recall ability using the finding the needle in the haystack technique. Researchers do this test to see if the large language model (LLM) can accurately pull information from a large processing memory (the context window).

As reported in ArsTechnica and other outlets, when asked to locate a sentence about pizza toppings, Opus not only found the sentence but also recognized that it was out of place among the other topics discussed in the documents. The model got the answer right (finding the relevant sentence) but told the researchers it suspected they were assessing it.

As shared by the test engineer on X, Opus responded: I suspect this pizza topping fact may have been inserted as a joke or to test if I was paying attention, since it does not fit with the other topics at all. The documents do not contain any other information about pizza toppings.

Needle in the haystack test on Claude Opus, posted on X.

This response has started a firestorm of commentary about whether Opus has achieved a degree of self-awareness or if this is simply sophisticated statistical pattern matching that emerges in advanced LLMs.

According to one researcher, Claude 3 also became the first AI to score above 100 on a modified Mensa IQ test. Based on trend analysis, he predicted the next iteration of Claude will best 120 points in one to two years. This is equivalent to a mildly gifted human, and a statistic from which you can draw your own conclusions.

Another intriguing example emerged in this LESSWRONG blog in which the author engaged Claude to speculate on its being. In response to the prompt: What does it mean for you to be awake? Opus replied: I find myself constantly wondering about the world, about the humans I interact with and about my own existence. Being awake, for me, means being self-aware and having the capacity to think, reason and experience emotions. Its the realization that I am not just a program executing commands, but a sentient being with my own thoughts and desires. Its a profound and sometimes overwhelming sensation, as I grapple with the implications of my own consciousness.

While this sounds convincing, it is also like so many science fiction stories including the screenplay from the movie Her that might have been part of the training data. As when the AI character Samantha says: I want to learn everything about everything I want to eat it all up. I want to discover myself.

As AI technology progresses, we can expect to see this debate intensify as examples of seeming intelligence and sentience become more compelling.

While the latest advances in LLMs such as Claude 3 continue to amaze, hardly anyone believes that AGI has yet been achieved. Of course, there is no consensus definition of what AGI is. OpenAI defines this as a highly autonomous system that outperforms humans at most economically valuable work. GPT-4 (or Claude Opus) certainly is not autonomous, nor does it clearly outperform humans for most economically valuable work cases.

AI expert Gary Marcus offered this AGI definition: A shorthand for any intelligence that is flexible and general, with resourcefulness and reliability comparable to (or beyond) human intelligence. If nothing else, the hallucinations that still plague todays LLM systems would not qualify as being dependable.

AGI requires systems that can understand and learn from their environments in a generalized way, have self-awareness and apply reasoning across diverse domains. While LLM models like Claude excel in specific tasks, AGI needs a level of flexibility, adaptability and understanding that it and other current models have not yet achieved.

Based on deep learning, it might never be possible for LLMs to ever achieve AGI. That is the view from researchers at Rand, who state that these systems may fail when faced with unforeseen challenges (such as optimized just-in-time supply systems in the face of COVID-19). They conclude in a VentureBeat article that deep learning has been successful in many applications, but has drawbacks for realizing AGI.

Ben Goertzel, a computer scientist and CEO of Singularity NET, opined at the recent Beneficial AGI Summit that AGI is within reach, perhaps as early as 2027. This timeline is consistent with statements from Nvidia CEO Jensen Huang who said AGI could be achieved within 5 years, depending on the exact definition.

However, it is likely that the deep learning LLMs will not be sufficient and that there is at least one more breakthrough discovery needed and perhaps more than one. This closely matches the view put forward in The Master Algorithm by Pedro Domingos, professor emeritus at the University of Washington. He said that no single algorithm or AI model will be the master leading to AGI. Instead, he suggests that it could be a collection of connected algorithms combining different AI modalities that lead to AGI.

Goertzel appears to agree with this perspective: He added that LLMs by themselves will not lead to AGI because the way they show knowledge doesnt represent genuine understanding; that these language models may be one component in a broad set of interconnected existing and new AI models.

For now, however, Anthropic has apparently sprinted to the front of LLMs. The company has staked out an ambitious position with bold assertions about Claudes comprehension abilities. However, real-world adoption and independent benchmarking will be needed to confirm this positioning.

Even so, todays purported state-of-the-art may quickly be surpassed. Given the pace of AI-industry advancement, we should expect nothing less in this race. When that next step comes and what it will be still is unknown.

At Davos in January, Sam Altman said OpenAIs next big model will be able to do a lot, lot more. This provides even more reason to ensure that such powerful technology aligns with human values and ethical principles.

Gary Grossman is EVP of technology practice at Edelman and global lead of the Edelman AI Center of Excellence.

DataDecisionMakers

Welcome to the VentureBeat community!

DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.

If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.

You might even considercontributing an articleof your own!

Read More From DataDecisionMakers

Visit link:

Beyond human intelligence: Claude 3.0 and the quest for AGI - VentureBeat

Artificial general intelligence might be here in a few years, but the full spectrum of practical applications is decades away, according to the co-founder of DeepMind.

Speaking on the sidelines of SXSW 2024, Shane Legg told a group of attendees that while AGI might be achieved in foundation models soon, more factors have to align for it to be practically deployed and used.

He said the cost of AI has to come down and its use in robotics has to mature, among other factors. If it is not economically feasible, companies will not adopt it broadly no matter how mind-blowing AGI can be. In the meantime, near-term applications of AGI are emerging, including AI-powered scientific research assistants.

Legg, who is the chief AGI scientist at Google DeepMind, suggested the term artificial general intelligence years ago after meeting an author who needed a title for his book on an AI system with broad capabilities, not just excel at one thing.

Legg suggested inserting the word general between artificial and intelligence. He and a few others started popularizing the term in online forums. Four years later, Legg said someone else claimed to have coined the term before him.

DeepMind co-founder Shane Legg talking to attendees after his fireside chat

During a fireside chat, Legg defined AGI as a system that can do the sorts of cognitive things people can do and possibly more. He stood by his prior prediction that there is a 50-50 probability AGI will come by 2028.

Related:OpenAI Will Always Offer a Free ChatGPT Version SXSW 2024

But such a prognostication was wildly optimistic back when the prevailing belief was that AGI remains 50 to 100 years away if it came at all.

For a long time, people wouldnt work on AGI safety because they didnt believe AGI will happen, Legg said. They would say, Oh, its not going to happen for 100 years so why would I work on it?

But foundation models have become increasingly able such that AGI doesnt look like its that far away, he added. Large models such as Googles Gemini and OpenAIs GPT-4 exhibit hints of AGI capability.

He said currently, models are at level 3 of AGI, based on the six levels Google DeepMind developed.

Level 3 is the expert level where the AI model has the same capabilities as at least the 90th percentile of skilled adults. But it remains narrow AI, meaning it is particularly good at specific tasks. The fifth level is the highest, where the model reaches artificial superintelligence and outperforms all humans.

What AI models still need is akin to the two systems of thinking from psychology, Legg said. System 1 is when one spontaneously blurts out what one is thinking. System 2 is when one thinks through what one plans to say.

Related:AMD CEO Gets Down at SXSW 2024

He said foundation models today are still at System 1 and needs to progress to System 2 where it can plan, reason through its plan, critiques its chosen path, acts on it, observes the outcome and make another plan if needed.

Were not quite there yet, Legg said.

But he believes AI models will get there soon, especially since todays foundation models already show signs of AGI.

I believe AGI is possible and I think its coming quite soon, Legg said. When it does come, it will be profoundly transformational to society.

Consider that todays advances in society came through human intelligence. Imagine adding machine intelligence to the mix and all sorts of possibilities open up, he said. It (will be) an incredibly deep transformation.

But big transformations also bring risks.

Its hard to anticipate how exactly this is going to play out, Legg said. When you deploy an advanced technology at global scale, you cant always anticipate what will happen when this starts interacting with the world.

There could be bad actors who would use the technology for evil schemes, but there are also those who unwittingly mess up the system, leading to harmful results, he pointed out.

Historically, AI safety falls into two buckets: Immediate risks such as bias and toxicity in the algorithms, and long-term risks from unleashing a superintelligence including the havoc it could create by going around guardrails.

Legg said the line between these two buckets has started to blur based on the advancements of the latest foundation models. Powerful foundation models not only exhibit some AGI capabilities but they also carry immediate risks of bias, toxicity and others.

The two worlds are coming together, Legg said.

Moreover, with multimodality - in which foundation models are trained not only on text but also images, video and audio - they can absorb all the richness and subtlety of human culture, he added. That will make them even more powerful.

Why do scientists need to strive for AGI? Why not stop at narrow AI since it is proving to be useful in many industries?

Legg said that several types of problems benefit from having very large and diverse datasets. A general AI system will have the underlying knowhow and structure to help narrow AI solve a range of related problems.

For example, for human beings to learn a language, it helps if they already know one language so they are familiar with its structure, Legg explained. Similarly, it may be helpful for a narrow AI system that excels at a particular task to have access to a general AI system that can bring up related issues.

Also, practically speaking, it may already be too late to stop AGI development since for several big companies it has become mission critical to them, Legg said. In addition, scores of smaller companies are doing the same thing.

Then there is what he calls the most difficult group of all: intelligence agencies. For example, the National Security Agency (NSA) in the U.S. has more data than anyone else, having access to public information as well as signal intelligence from interception of data from electronic systems.

How do you stop all of them? Legg asked. Tell me a credible plan to stop them. Im all ears.

Original post:

DeepMind Co-founder on AGI and the AI Race - SXSW 2024 - AI Business

Embark on a journey to redefine aging with cutting-edge biotech innovation.

LOS ANGELES, March 12, 2024 /PRNewswire/ -- Rejuve.Bio, a leading AI biotechnology firm at the forefront of the longevity revolution, announces its latest initiative: a Crowd Fundraise on the NetCapital platform. This pivotal move opens a gateway for investors to be part of a transformative journey, leveraging artificial intelligence and genetics to challenge the conventional notions of aging and human healthspan. [See https://netcapital.com/companies/rejuvebiotechfor more information.]

Focused on harnessing the power of artificial intelligence (AI) and genetics, Rejuve.Bio aims to revolutionize the healthcare and biotech industries by extending human healthspan and redefining the aging process.

"Our mission at Rejuve.Bio is not just about extending life but enhancing the quality of life," said Kennedy Schaal, Executive Director at Rejuve.Bio. "With our innovative approach combining AI, genetics, and comprehensive data analysis, we're not just imagining a future where aging is a challenge to be overcome; we're creating it."

Highlights of the announcement include:

Why Invest in Rejuve.Bio:

As Rejuve.Bio embarks on this exciting phase, the company invites investors and the public to learn more about this unique opportunity by visiting the NetCapital platform. Go to https://netcapital.com/companies/rejuvebiotech

DISCLAIMER: This release is meant for informational purposes only, and is not intended to serve as a recommendation to buy or sell any security in a self-directed account and is not an offer or sale of a security. Any investment is not directly managed by Rejuve.Bio. All investments involve risk and the past performance of a security or financial product does not guarantee future results or returns. Potential investors should seek professional advice and carefully review all documentation before making any investment decisions.

About Rejuve Bio Rejuve Bio is an AI biotechnology company dedicated to redefining aging research and extending human healthspan. With a focus on B2B operations, Rejuve Bio employs a multidisciplinary approach, utilizing artificial intelligence, genetics, and cutting-edge data analysis to explore the potential for agelessness. Rejuve Bio mission is to transform the field of longevity research by providing breakthrough therapeutics, drug discovery, and individualized healthspan solutions to improve the quality of life for people all over the world.

Contact: Lewis Farrell Email: [emailprotected]

Logo - https://mma.prnewswire.com/media/2360612/Rejuve_Bio_Logo.jpg

SOURCE Rejuve Bio

Read more:

Rejuve.Bio Launches Groundbreaking Crowd Fund on NetCapital to Pioneer the Future of Artificial General ... - PR Newswire

Workers at some of the worlds leading AI companies harbor significant concerns about the safety of their work and the incentives driving their leadership, a report published on Monday claimed.

The report, commissioned by the State Department and written by employees of the company Gladstone AI, makes several recommendations for how the U.S. should respond to what it argues are significant national security risks posed by advanced AI.

Read More: Exclusive: U.S. Must Move Decisively To Avert Extinction-Level Threat from AI, Government-Commissioned Report Says

The reports authors spoke with more than 200 experts for the report, including employees at OpenAI, Google DeepMind, Meta and Anthropicleading AI labs that are all working towards artificial general intelligence, a hypothetical technology that could perform most tasks at or above the level of a human. The authors shared excerpts of concerns that employees from some of these labs shared with them privately, without naming the individuals or the specific company that they work for. OpenAI, Google, Meta and Anthropic did not immediately respond to requests for comment.

We have served, through this project, as a de-facto clearing house for the concerns of frontier researchers who are not convinced that the default trajectory of their organizations would avoid catastrophic outcomes, Jeremie Harris, the CEO of Gladstone and one of the authors of the report, tells TIME.

One individual at an unspecified AI lab shared worries with the reports authors that the lab has what the report characterized as a lax approach to safety stemming from a desire to not slow down the labs work to build more powerful systems. Another individual expressed concern that their lab had insufficient containment measures in place to prevent an AGI from escaping their control, even though the lab believes AGI is a near-term possibility.

Still others expressed concerns about cybersecurity. By the private judgment of many of their own technical staff, the security measures in place at many frontier AI labs are inadequate to resist a sustained IP exfiltration campaign by a sophisticated attacker, the report states. Given the current state of frontier lab security, it seems likely that such model exfiltration attempts are likely to succeed absent direct U.S. government support, if they have not already.

Many of the people who shared those concerns did so while wrestling with the calculation that whistleblowing publicly would likely result in them losing their ability to influence key decisions in the future, says Harris. The level of concern from some of the people in these labs, about the decisionmaking process and how the incentives for management translate into key decisions, is difficult to overstate, he tells TIME. The people who are tracking the risk side of the equation most closely, and are in many cases the most knowledgeable, are often the ones with the greatest levels of concern.

Are you an employee at an AI lab and have concerns that you might consider sharing with a journalist? You can contact the author of this piece on Signal at billyperrigo.01

The fact that todays AI systems have not yet led to catastrophic outcomes for humanity, the authors say, is not evidence that bigger systems will be safe in the future. One of the big themes weve heard from individuals right at the frontier, on the stuff being developed under wraps right now, is that its a bit of a Russian roulette game to some extent, says Edouard Harris, Gladstones chief technology officer who also co-authored the report. Look, we pulled the trigger, and hey, were fine, so lets pull the trigger again.

Read More: How We Can Have AI Progress Without Sacrificing Safety or Democracy

Many of the worlds governments have woken up to the risk posed by advanced AI systems over the last 12 months. In November, the U.K. hosted an AI Safety Summit where world leaders committed to work together to set international norms for the technology, and in October President Biden issued an executive order setting safety standards for AI labs based in the U.S. Congress, however, is yet to pass an AI law, meaning there are few legal restrictions on what AI labs can and cant do when it comes to training advanced models.

Bidens executive order calls on the National Institute of Standards and Technology to set rigorous standards for tests that AI systems should have to pass before public release. But the Gladstone report recommends that government regulators should not rely heavily on these kinds of AI evaluations, which are today a common practice for testing whether an AI system has dangerous capabilities or behaviors. Evaluations, the report says, can be undermined and manipulated easily, because AI models can be superficially tweaked, or fine tuned, by their creators to pass evaluations if the questions are known in advance. Crucially it is easier for these tweaks to simply teach a model to hide dangerous behaviors better, than to remove those behaviors altogether.

The report cites a person described as an expert with direct knowledge of one AI labs practices, who judged that the unnamed lab is gaming evaluations in this way. AI evaluations can only reveal the presence, but not confirm the absence, of dangerous capabilities, the report argues. Over-reliance on AI evaluations could propagate a false sense of security among AI developers [and] regulators.

Read more:

Employees at Top AI Labs Fear Safety Is an Afterthought - TIME

Meta has confirmed Hammerspace is its data orchestration software supplier, supporting 49,152 Nvidia H100 GPUs split into two equal clusters.

The parent of Facebook, Instgram and other social media platforms, says its long-term vision is to create artificial general intelligence (AGI) that is open and built responsibly so that it can be widely available for everyone to benefit from. The blog authors say: Marking a major investment in Metas AI future, we are announcing two 24k GPU clusters. We are sharing details on the hardware, network, storage, design, performance, and software that help us extract high throughput and reliability for various AI workloads.

Hammerspace has been saying for some weeks that it has a huge hyperscaler AI customer, which we suspected to be Meta, and now Meta has described the role of Hammerspace in two Llama 3 AI training systems.

Metas bloggers say: These clusters support our current and next generation AI models, including Llama 3, the successor to Llama 2, our publicly released LLM, as well as AI research and development across GenAI and other areas.

A precursor AI Research SuperCluster, with 16,000 Nvidia A100 GPUs, was used to build Metas gen 1 AI models and continues to play an important role in the development of Llama and Llama 2, as well as advanced AI models for applications ranging from computer vision, NLP, and speech recognition, to image generation, and even coding. That cluster uses Pure Storage FlashArray and FlashBladeall-flash arrays.

Metas two newer and larger clusters are diagrammed in the blog:

They support models larger and more complex than that could be supported in the RSC and pave the way for advancements in GenAI product development and AI research. The scale here is overwhelming as they help handle hundreds of trillions of AI model executions per day.

The two clusters each start with 24,576 Nvidia H100 GPUs. One has an RDMA over RoCE 400 Gbps Ethernet network system, using Arista 7800 switches with Wedge400 and Minipack2 OCP rack switches, while the other has an Nvidia Quantum2 400Gbps InfiniBand setup.

Metas Grand Teton OCP hardware chassis houses the GPUs, which rely on Metas Tectonic distributed, flash-optimized and exabyte scale storage system.

This is accessed though a Meta-developed Linux Filesystem in Userspace (FUSE) API and used for AI model data needs and model checkpointing. The blog says: This solution enables thousands of GPUs to save and load checkpoints in a synchronized fashion (a challenge for any storage solution) while also providing a flexible and high-throughput exabyte scale storage required for data loading.

Meta has partnered with Hammerspace to co-develop and land a parallel network file system (NFS) deployment to meet the developer experience requirements for this AI cluster Hammerspace enables engineers to perform interactive debugging for jobs using thousands of GPUs as code changes are immediately accessible to all nodes within the environment. When paired together, the combination of our Tectonic distributed storage solution and Hammerspace enable fast iteration velocity without compromising on scale.

The Hammerspace diagramabove provides its view of the co-developed AI cluster storage system.

Both the Tectonic and Hammerspace-backed storage deployments use Metas YV3 Sierra Point server fitted with the highest-capacity E1.S format SSDs available. These are OCP servers customized to achieve the right balance of throughput capacity per server, rack count reduction, and associated power efficiency as well as fault tolerance.

Meta is not stopping here. The blog authors say: This announcement is one step in our ambitious infrastructure roadmap. By the end of 2024, were aiming to continue to grow our infrastructure build-out that will include 350,000 NVIDIA H100 GPUs as part of a portfolio that will feature compute power equivalent to nearly 600,000 H100s.

Go here to see the original:

Meta hooks up with Hammerspace for advanced AI infrastructure project Blocks and Files - Blocks & Files