Рубрика: AI

Meta has kept up in the AI race, releasing new generative AI models, features, and research constantly. However, the company is now positioning itself to dive into artificial general intelligence (AGI).

On Thursday, Meta CEO Mark Zuckerberg took to Threads to post an almost two-minute video updating the public on the company's AI efforts, which included a glimpse into its long-term goals for building AGI.

"Today I'm bringing Meta's two AI research efforts [FAIR and GenAI] closer together to support our long-term goals of building general intelligence, open-sourcing it responsibly, and making it available and useful to everyone in all of our daily lives," said Zuckerberg in the opening of the video.

Also: ChatGPT goes to college: OpenAI finds its first higher education partner

He also added that the need for building total AGI has become more prevalent than ever since the technology will be crucial to fueling the next generation of AI services and offering the best AI-enabled assistance.

As a result, Meta is building a massive amount of infrastructure. By the end of the year, the company will have 350,000 Nvidia H100s, or around 600,000 H100s equivalents of compute if you include other GPUs.

Also: Five ways to use AI responsibly

To put that compute power in perspective, Meta and Microsoft were the largest purchasers of Nvidia's H100 GPUs in 2023, procuring as many as 150,000 each, according to analyst firm Omdia. Meta's new goal of 350,000 H100s by the end of this year is more than double its already steep investment.

So what is AGI, and why is Meta so focused on investing and developing it?

Also: What is artificial general intelligence really about?

To put it simply, AGI refers to AI that can perform complex tasks, such as learning, reasoning, and more, just as well as humans. Currently, even the most advanced generative AI models still require human direction and instruction to complete a task.

AGI makes it possible to remove human instruction, direction, and supervision because the model can function autonomously. This development should significantly boost how AI can assist humans and even what tasks it can perform.

However, AGI's unprecedented intelligence and its ability to mimic human behavior are often the reasons people approach the technology with fear or apprehension. Until now, AGI has remained a long-term goal, but Meta's updates show the technology might be developed sooner than anticipated.

Also: The best AI chatbots: ChatGPT and other noteworthy alternatives

While wrapping up the video, Zuckerberg also shared that Meta is currently training LLaMa 3, which is the latest version of Meta's advanced large language model, LLaMa 2, and he teased the creation of other future models.

"Overall across all this stuff, we're just getting started," Zuckerberg concluded.

Artificial Intelligence

Tensoic, the creator of Kannada Llama, aka Kan-Llama, recently released the playground to test the model, which was in partnership with E2E Networks built on NVIDIA A100s infrastructure and with Xylem AI for inference.

Aditya Shirawalmath, the brain behind the model, said that Xylem AI’s inference platform has a mind-blowing inference stack.

AIM got in touch with Arko Chattopadhyay, the co-founder and CEO of Xylem AI, to understand the company’s vision and what they are offering to help build Indic LLMs, which makes them stand apart from others in the field, such as Together AI and Anyscale.

“We started as a search index company for the enterprise, something that Perplexity AI is doing for the web. We wanted to do it for enterprises’ own data,” said Chattopadhyay. The company pivoted its model three months into the business as it realised that there is not enough infrastructure to support building LLMs, and thus wanted to fill that space.

Along with Chattopadhyay, the nine-month-old startup was co-founded by Enrique Ferrao and Pranav Reddy. All three of them met at Manipal Institute of Technology when they started participating in hackathons. After founding Amigo, a mental healthcare services app for enterprise, the founders began working on Xylem AI.

No need for engineering efforts

Xylem offers a comprehensively managed LLMOps platform designed for teams to efficiently train, deploy, and scale LLMs in production, eliminating the need for additional engineering involvement. “We want to eliminate the need to hire expensive software engineers for companies to build LLMs,” said Chattopadhyay, emphasising that Xylem’s platform is very easily scalable for everyone.

Highlighting the privacy and security concerns that enterprises have with using closed-source models such as OpenAI’s GPT and Anthropic’s Claude, Chattopadhyay said that Xylem helps in fine-tuning with personal data on open source models, which increases the adoption by enterprises.

“This is also very compute-heavy for enterprises, which makes it very costly for companies,” he added. “We are building our fine-tuning stack that allows people to build their custom models fine-tuned much faster and cheaper. Our whole idea is that we don’t want to throw GPUs at scale, but optimise the models to run efficiently on GPUs,” said Chattopadhyay.

Xylem claims that a model fine-tuned on Llama 7B on its platform can deliver 110 tokens per second on a single inference and also enables 30 parallel requests on an NVIDIA A100 80 GB GPU. “We are trying to push that even more,” added Chattopadhyay. The company is also increasing the number of open source models available on the platform as well as the speed.

Currently, Xylem is scaling on a managed cloud of E2E Networks and running on NVIDIA GPUs, but it also has plans to expand to AMD’s latest offerings and is looking for more options, such as Elon Musk’s xAI’s custom build chip.

The goal is to be one of the fastest inference engine

“The battle to be the fastest inference engine is probably a zero-sum game,” said Chattopadhyay. “You can bring your data and train your model on our platform. We will take care of the backend and the engineers don’t have to waste any of their time,” he said.

He further added that once the model is trained, you can easily deploy it with Xylem’s platform, and keep updating the models as they come along.

Chattopadhyay claims that Xylem is currently on par with Together AI and Anyscale, and is very close to Perplexity AI when it comes to inference. “We don’t need to get any faster than this. LLMs would get better and the responses would be instantaneous for the end users,” he added. “It won’t matter,” he said that the conversation would shift to easier API integration and smoother developer experience.

“Building an LLM should not be the moat of the company, the data should be the moat,” said Chattopadhyay and added that anyone can build LLMs these days, but the point is how do you deliver it to the audience. “Since most of them are anyway building on top of other LLMs, there is no need to worry about deploying it to the people.”

“Of course, we are focusing on India as it is the biggest market and Indic LLMs are coming up, but our goal is to go global,” he said.

Chattopadhyay further said that the company already has customers in the Middle East and Europe, along with Indian companies, and is also in talks with companies such as Sarvam AI and Japanese AI startups Sakana AI.

“If you want to deploy a model for 1.5 billion people, you cannot just keep on stacking GPUs as the cost would be very high,” Chattopadhyay concluded, saying that not everyone can focus on every part of the stack of building and deploying LLMs.

That is where Xylem wants to take its place in the inference, fine-tuning, and deployment of the models for others, and focus on making the developer experience easier and faster.

The post How Xylem AI is Helping Build LLMs in India appeared first on Analytics India Magazine.

Walt Disney recently introduced HoloTiles, the world’s first multi-person, omnidirectional, modular, expandable treadmill floor. This new technology enables multiple users to share a VR experience, walk in any direction, and avoid collisions.

“It’s a very special piece of technology… I can walk on this omnidirectional floor in any direction I want. It will automatically do whatever it needs to have me stay on the floor. And what’s amazing about this is multiple people can be on it and all walking independently,” said Lanny Smoot, a seasoned Imaginer at Walt Disney Imagineering R&D, and the brain behind HoloTiles.

He envisions its application in creating immersive virtual reality experiences, transforming theatrical stages, and offering new possibilities in entertainment and beyond. He currently has a total of 106 patents. Some of the other notable inventions and contributions include Electromagnetic Eyes for Animatronices, Interactive Zoetrope, and Extendable Lightsaber.

Smoot joined Walt Disney in 1999 after working for over two decades in the telecommunication industry and at Bell Labs. Smoot has been recognised with induction into the National Inventors Hall of Fame. This prestigious honour places him alongside Walt Disney himself, marking him as only the second Disney employee to receive this distinction.

Walt Disney Imagineering

Walt Disney Imagineering, founded by Walt Disney himself, is renowned for blending creative imagination with technical know-how. It has a rich history of pioneering innovations in themed entertainment, starting from the creation of Disneyland in the 1950s.

Recently, Imagineering revealed Star Wars: Galaxy’s Edge, an immersive themed land at Disneyland and Disney’s Hollywood Studios. Additionally, it introduced Ant-Man and The Wasp: Nano Battle! at Hong Kong Disneyland, showcasing advanced interactive technologies.

Walt Disney Imagineering continues to push the boundaries of creativity and technology in theme park experiences and attractions. Check out all of their recent products here.

“I sometimes forget that Disney is actually a very, very good robotics hardware company,” said Jim Fan, senior research scientist and lead of AI agents at NVIDIA, pointing at a lifelike Shaman robot in the Na’vi River Journey (from movie Avatar) at Disneyland. “Such fluid & expressive motion,” he added.

“Note that this is animatronics – robots that don’t have autonomy but only follow fixed, artist-curated moves. If this level of hardware is combined with state-of-the-art robot foundation models, I’d imagine the potential would be off the charts,” said Fan.

The post Disney Unveils HoloTile, the World’s First Omnidirectional, Multi-User Treadmill appeared first on Analytics India Magazine.

If 2023 was the year of fearing generative AI, 2024 will be the year for some of those worries to come true.

Last summer, Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Yarin Gal, Nicolas Papernot, and Ross Anderson wrote a paper hinting at AI models poisoning themselves in the (near) future. The warning was seen as farsighted and only theoretical, but evidence of the problematic technology has emerged.

The problem called “model collapse”, where AI chatbots lose the information they learn, is not in theory any more. Last month, a Twitter user posted a screenshot showing that Grok, the large language model chatbot developed by Elon Musk’s AI company xAI, had plagiarised a response from OpenAI.

When asked by Winterbourne to tinker with malware, Grok responded that it could not, “as it goes against OpenAI’s use case policy”.

“Grok is literally just ripping OpenAI’s code base,” the user explained in the post. The explanation was denied by a technical staff member of xAI who has previously worked for rivals OpenAI and Google DeepMind.

“This was a huge surprise to us when we first noticed it,” he responded. The staff member might not have seen this coming, but the company’s chief, Musk, definitely did.

Following the screenshot, which amassed a lot of reactions, ChatGPT shared the same and wrote, “We have a lot in common.” Musk noted the same and responded, “Well, son, since you scraped all the data from this platform for your training, you ought to know.”

The technology has given rise to competition not just among tech companies but rehashed old rivalries like the one between OpenAI and Musk, who was a cheerleader of the GPT maker earlier.

Leaving tech bros’ personal problems aside, AI-related error messages have also entered online shopping lists. Users on the e-commerce platform Amazon have pointed out that OpenAI error messages appear in products, be it lawn chairs or Chinese religious tracts.

The original copies of these products are named “I’m sorry, but I cannot fulfil this request. It goes against OpenAI use policy” have been archived after media publications discovered the listings. Still, many such artificial posts can be found on Threads and LinkedIn.

Delusions, Delusions, Delusions

Many said that the research by Shumailov at the University of Oxford and his colleagues overlooked an essential point. One of them was Daniel Sack, managing director and partner at Boston Consulting Group X (the tech build and design unit of BCG).

“Most of the data that will be used to train future models will not be mere reproductions of the source material but entirely novel and unprecedented,” he wrote on LinkedIn.

His theory, in response, is understandable since people in tech usually have a hard time calling out the mishaps of the products they are building or helping someone else build. Time and again, Silicon Valley has hesitated to acknowledge the menace of unwanted technologies.

The curious case of generative AI models is even harder to pinpoint since a lot of money is riding on the game.

Even Sack’s firm, BCG X, has collaborated with OpenAI, revealing that none in favour of the technology can be trusted, at least for now, since it has layers of unsolved ethical issues. All the above issues show that boasting about the technology’s capabilities to solve humanity’s grave problems should not be the primary response.

No Way Back

Generative AI programs rely on unfathomable amounts of data from every nook and cranny of the internet. The web has already become awash with AI-generated spam. No matter how much the VCs or developers of these AI models deny, the problem exists and is only going to get worse from here as hundreds of millions of people are using these tools every day.

“It really shows that these models are not going to be reliable in the long run if they learn from post-LLM age data—without being able to tell what data has been machine-generated, the quality of the outputs will continue to decline,” Catherine Flick, a professor of ethics and games technology at Staffordshire University told Fast Company while speaking about the Grok incident.

Foremostly, there is no way for humans to differentiate between AI-generated and human-generated content. Similarly, these language models also have no way of telling whether the AI-generated text they see corresponds to reality, which could introduce even more misinformation than current models.

For now, all one can do is sit back and watch the internet burn.

The post How Generative AI is Gobbling Up the Internet appeared first on Analytics India Magazine.

AI has arrived, in full force, on phones for the biggest players on the market, and, with the Galaxy S24, Samsung is coming to prove it can stand toe-to-toe with Google and Apple. The new Galaxy phones are available for preorders starting today, in sync with the company's Unpacked event.

Also: Every product Samsung announced at Unpacked 2024

This year, we're seeing some serious upgrades (a new Qualcomm chip, a brighter screen, and an even further refined design). But as far as the new capabilities for the phones, AI takes the spotlight, thanks to Galaxy AI.

Samsung Unpacked

Meta chief Zuckerberg has thrown his hat into the ring in the pursuit of AGI. In his recent Instagram post, he announced, “Our long-term vision is to build general intelligence, open-source it responsibly, and make it widely available so everyone can benefit.”

To achieve this, he merged his two major AI research efforts, FAIR and the GenAI team. “We’re bringing our two major AI research efforts (FAIR and GenAI) closer together to support this,” his post read.

The merger of the FAIR and GenAI teams is truly intriguing. FAIR is dedicated to research, while the GenAI team is focused on generative AI experiences for users on Meta’s apps. During last year’s Meta Connect, new features for creators were announced on Facebook and Instagram, which included AI-powered image editing, sticker generation, and personalised recommendations.

This merger reminded many of the Google Brain and DeepMind merger that took place last year. “FAIR was part of Reality Labs – Research (RL-R), primarily focusing on the Metaverse, AR, VR, and MR. Given the increasing importance of AI and FAIR’s close relationship with GenAI, it made sense for FAIR and GenAI to be under the same umbrella,” wrote Yann Lecun, Meta’s AI chief scientist.

GPUs = AGI?

Considering Meta’s computational resources, they might surge ahead in the AGI race. “We’re currently training our next-gen model Llama 3, and we’re building massive compute infrastructure to support our future roadmap, including 350k H100s by the end of this year — and overall almost 600k H100s equivalents of compute if you include others” said Zuckerberg.

The current number of GPUs possessed by Meta doesn’t come as a surprise, given that last year, Meta had 150K H100s, the highest compared to other players, including Google, Amazon, and Oracle.

Interestingly, this marks the first time a major tech company has openly disclosed precise figures. On the flip side, the number of GPUs at OpenAI remains undisclosed, although Sam Altman hinted that they possess enough for training GPT-5.

While Zuckerberg has set his sights on AGI, Sam Altman recently downplayed its impact and said, ‘It will change the world much less than we all think, and it will change jobs much less than we all think,’ during a conversation at the World Economic Forum in Davos, Switzerland.

“I believe that someday we will make something that qualifies as an AGI by whatever fuzzy definition you want, the world will have a two-week freakout and then people will go on with their lives,” he expressed in another conversation at Davos .

AGI in Metaverse

Zuckerberg is hopeful that once AGI is achieved, it will not exist solely in the physical world but rather in a blend of virtual and physical reality. “People are also going to need new devices for AI and the Metaverse. Because over time I think a lot of us are going to talk to AIs frequently throughout the day.” he said.

Further, he opines that Meta’s Ray-Ban glasses are the ideal form factor for enabling AI to see what you see and hear what you hear. “Ray Ban Meta Glasses with Meta AI are off to a very strong start and overall across all this stuff we are just getting started”, he said.

Additionally, he dropped hints that Meta is still committed to the Metaverse and actively investing in it. The company is allocating an annual budget of over $15 billion to support Reality Labs and advance the development of the metaverse. Last year, Zuckerberg appeared on a podcast with Lex Fridman, which also happened to be the first interview to happen in the Metaverse.

Is Meta the new OpenAI?

Ultimately, the concept of open-source AGI might compel OpenAI to reconsider its strategy, given that the company initially embraced the idea of open-sourcing models.

At the recently concluded World Economic Forum, LeCun advocated for open-source foundational models emphasising that OpenAI wouldn’t be what it is today without the contributions from the open-source community.

“OpenAI does not have a monopoly on good ideas. They’re not going to get to AGI by themselves; in fact, they’re using PyTorch and Transformers, which were published by many of us. They’re profiting from the open research landscape,” he said.

“Zuck and Yann LeCun will go down as heroes in human history! Fighting for ‘Open AI’ when the incumbents sought to shut it down! Unbelievable how much the vibes from Meta have changed over the last year. That, and maybe it’s time for a name change – Meta to OpenAI,” wrote Bindu Reddy, Chief at Abacus.ai, on X.

Likewise, Perplexity chief Arvind Srinivas said,” Open Source AGI is an amazing vision. You (Meta) are building a very powerful technology, and actually aligning to what makes sense for the world: more people have a say in what makes sense and doesn’t”.

The post Meta Eyes AGI with Llama 3 appeared first on Analytics India Magazine.

Every time some astrologer, shaman, or random person in the street tells me the world will end on a certain date, I become suffused with skepticism.

If I knew exactly when the world was going to end, I'm not sure I'd walk the streets and tell everyone else. Instead, I'd go off and have the finest experiences I could before the Earth's bouncers shut the door for the final time.

Also: 4 ways to overcome your biggest worries about generative AI

Yet here we are, undergoing the sudden onrush of AI, and predictions of imminent doom are all around.

In its enthusiasm, OpenAI — makers of ChatGPT — once fueled the frenzy. Why, it was only last October that OpenAI's then-CEO Sam Altman mused that AI might well kill us.

"Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war," he offered.

This wasn't precisely what those already frightened by the prospect of robot rule wanted to hear. It only fueled their fears and weighed down their hearts.

Talking it up, toning it down

Recently, however, Altman — once again the OpenAI CEO — seems to have tempered his tone. Even more happily, he seems to be taking steps to ensure not only our survival but also (what's left of) our peace of mind.

Why, last week he announced that OpenAI will take steps to make sure that the upcoming US elections won't be interfered with — by nefarious types using AI to achieve their own twisted ends.

But it was another of his pronouncements that may incite a certain giddiness in doubting, troubled hearts.

Also: How OpenAI plans to help protect elections from AI-generated mischief

Speaking at the World Economic Forum, Altman embraced an odd sanguinity. He said that AI "will change the world much less than we all think and it will change jobs much less than we all think."

I'm not sure we all think the same things. Why, every time I wake up and cast my bleary eyes upon the news, I often see words and gestures from people who don't think like me at all.

Still, Altman seems to have embraced many people's unease about not only what AI might do, but how quickly it might change, or even wreck, their lives.

And he wants you to know it may not be so dramatic.

Embracing your unease

Casting my innate skepticism to the floor for a moment, I'm moved by Altman's recent public utterances. I'm moved that they differ so much from the ones offered by tech leaders during the first part of the internet era.

Then, the likes of Google and Facebook just wanted everyone out of their way because they were "making the world a better place."

Because they said so. And, well, they were far too clever for ordinary people to understand what they were doing.

Also: This is why AI-powered misinformation is the top global risk

By contrast, Altman at least expresses awareness that hype and hysteria aren't necessarily the right notes for today's ever-shifting opera.

Of course, this might all have a little to do with the prospect of government regulation. Calm your predictions and you seem more reasonable.

It might also have something to do with the reasons Altman was briefly removed from OpenAI by those who were a little less keen on uncontrolled moneymaking.

One can still, though, appreciate his effort and even be slightly startled. Even if, in a subsequent interview last week, he conceded that, as AI becomes more customizable, "that's going to make a lot of people uncomfortable."

Also: AI will have a big impact on jobs this year. Here's why that could be good news

Altman concedes that AI is already an "incredible tool for productivity." Having used it quite a bit myself — no, not for this column, silly — I concede that its productivity is extraordinary.

However, its productivity in video is far more impressive — and mind-challenging — than its writing ability. Lordy, does ChatGPT write like the fifth-grade-level robot science project of a reclusive, asocial professor!

Come out of your bunker, Archie

As the hype abates, perhaps the practical uses of AI will be judged on their merits. Even the fearful, the creeped-out, and the skeptical might become more calmly aware of AI's presence, its benefits, and its manifest deficiencies.

We're headed for interesting times — yes, even more of them — but it's worth emerging from our bunkers and being interested in what's occurring all around us without being excessively frightened.

No, that wasn't a political statement, but I thought I'd see what ChatGPT thought.

Also: Afraid of AI? I confronted it for you and its responses were fascinating

"Is it worth emerging from our bunkers and being interested in what's occurring all around us?" I asked.

"While I don't have personal opinions, I can provide some perspective," began the AI. "Staying informed about what's happening around you can be important for various reasons. It allows you to make informed decisions, engage in meaningful conversations, and be aware of changes that may impact your life."

But then it did offer some personal opinions: "Being knowledgeable about current events can also contribute to a sense of community and understanding of the world."

Ah, that sense of community around current events. Wouldn't it be wonderful if AI could bring us a little more of that?

Artificial Intelligence