How Meta wants to shape our digital future with open source AI w/ Ragavan Srinivasan

Computers have been outperforming humans for years on tasks like solving complex equations or analyzing data, but when it comes to the physical world, robots struggle to keep up. It can take years to train robots to function in the messy chaos of the “real world” — but thanks to some unlikely help from the film and video gaming industry, robots today are using AI to fast-track their learning and master new skills using simulated environments. Rev Lebaredian is the vice president of Omniverse and simulation technology at NVIDIA, a company known for its work on advancements in AI, video game graphics cards, and accelerated computing and computer graphics. Rev and Bilawal discuss how simulated “mirror worlds” can help robots learn faster, the trillion-dollar market for physical AI, and the future of AI robot assistance in our everyday lives.  For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts  

It’s impossible to separate the way people engage with AI with the way they engage with the  internet as a whole. This is an episode of ReThinking, another podcast from the TED Audio Collective, featuring a compelling discussion of why the internet can feel so unfriendly–and where we can go from there. You’ll hear from Jay Van Bavel, an award-winning professor of psychology and neural science at NYU, on the science of virality, why bad news commands our attention, and how we can find common ground around more uplifting content. If you liked this episode, you can find more ReThinking wherever you get your podcasts.

Ever since generative AI tools like Midjourney became available to the public in 2022, curious users and AI fanatics alike have been experimenting with the technology. But for tech aficionados and AI enthusiasts like Justin Maier and Maxfield Hulker, Midjourney’s closed-source model wasn’t enough — they wanted to go deeper. That’s why Justin and Max created Citivai, an open-source generative AI tool and social platform where users can create, share, and experiment with new image generation models. They sit down with Bilawal to discuss why community is so important to open-source development, the future of algorithmic personalization, and the famous so-called “dead internet theory.” They also unpack the risks of open-source development, and emphasize the importance of setting boundaries to keep users safe — while acknowledging the important role that “not-safe-for-work” content has played in the evolution of these powerful tools.For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts   

When it comes to preparing for an interview or making an important life decision, more and more people are turning to AI for advice. ChatGPT’s new voice interface, Advanced Voice Mode, allows users to speak out loud and converse with a chatbot as they would with another human — but is it really as seamless as a chat with a friend? Bilawal runs a series of experiments with Advanced Voice Mode to test the limits of this new technology and its potential uses, from weighing the pros and cons of a cross-country move to coaching an intense personal workout. He and producer Dominic Girard discuss the potential benefits and dangers of this new advancement, and ask perhaps the most important question of all: can ChatGPT pronounce Bilawal’s name?  For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts   

We’ve had conversations about AI’s online influence on politics, from deepfakes to misinformation. But AI can also have profound effects on hardware – especially when it comes to national security and military capabilities like weapons and stealth technologies. Kathleen Fisher is an office director at DARPA, the Defense Advanced Research Projects Agency tasked with the research and development of emerging technologies for use by the U.S. military. Despite its bureaucratic name, DARPA is anything but conventional – and they’re solving problems that are thrillingly complex. Kathleen shares how her team employs nimble thinking to understand the state of AI across the globe. Then, she and Bilawal discuss the strategies needed to embrace the possibilities –and challenges– of AI now, and what we need to do to build a sustainable future.For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts 

Many people associate innovation with secrecy—privately toiling away on a project until you're ready to share it with the world. While that may work for some, there's a benefit to putting all your cards on the table. Bilawal sits down for a conversation with Thomas Wolf, whose company Hugging Face pivoted from privately building an AI chatbot to sharing all of its knowledge with a growing online community. Thomas discusses the history of Hugging Face, why embracing open source development has shifted the trajectory of AI, and how open source can challenge existing power structures in the AI world.

From vetting resumes to screening candidates, many employers are using AI tools to identify top talent. But what happens when companies start relying on AI to help them decide who to hire or promote…and who to fire? Bilawal speaks with journalist Hilke Schellmann, whose research on the rapidly growing use of AI in the workplace highlights where algorithms are helping – and hurting – business. Hilke shares the surprising (and not surprising) ways AI works in the hiring process, and argues that transparency, regulation, and oversight are essential if AI is going to actually benefit employees and employers. For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts  

Llama is Meta’s Large Language Model trained on over 15 trillion tokens of publicly available information. It’s available to anyone – from people making custom fan-made entertainment on a smartphone… to, potentially, complex projects that may not have the public’s well-being in mind. So if Llama is such a widely available and powerful product, why is Meta releasing it – for free? Bilawal chats with Meta’s own Vice President of Product, Ragavan Srinivasan, to discuss the pressing questions around Llama’s benefits and risks.For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts  

Whether finding a restaurant or fact-checking a new claim, search engines are one of the main avenues we use to navigate the world. So why are modern engines so clunky and frustrating – and how is AI already changing the infrastructure we use to access information on the internet? Bilawal sits down with CEO of Perplexity AI Aravind Srinivas to discuss how we got to a world with too many links, and what the future may hold for your experience on the web.

If you've ever cringed at a poorly-dubbed film, you are not alone. That's why Scott Mann founded Flawless, a company that’s transforming the world of dubbing using AI. He talks with Bilawal about why good dubbing is essential for movie making and shares the mind-blowing technology that not only lets Robert DeNiro speak perfect Spanish, but radically changes how films might get made. The two also discuss what to keep in mind as creativity, industry, and AI technology continue to intertwine — and what we need to protect artists' rights.For transcripts for The TED AI Show, visit go.ted.com/TTAIS-transcripts