“This moment in AI is an inflection moment,” Fei-Fei Li told me recently. Li is co-director of Stanford’s Human-Centered AI Institute and some of the distinguished computer science researchers of our time. She is best known for creating ImageNet, a preferred image data set that was pivotal in allowing researchers to coach modern AI systems.
Two things have happened, Li explains. Generative AI has caused the general public to wake as much as AI technology, she says, since it’s behind concrete tools, resembling ChatGPT, that folks can check out for themselves. And because of this, businesses have realized that AI technology resembling text generation could make them money, and so they have began rolling these technologies out in additional products for the true world. “Due to that, it impacts our world in a more profound way,” Li says.
Li is one in every of the tech leaders we interviewed for the most recent issue of MIT Technology Review, dedicated to the largest questions and hardest problems facing the world. We asked big thinkers of their fields to weigh in on the underserved issues on the intersection of technology and society. Read what other tech luminaries and AI heavyweights, resembling Bill Gates, Yoshua Bengio, Andrew Ng, Joelle Pineau, Emily Bender, and Meredith Broussard, needed to say here.
In her newly published memoir, Li recounts how she went from an immigrant living in poverty to the AI heavyweight she is today. It’s a touching look into the sacrifices immigrants need to make to realize their dreams, and an insider’s telling of how artificial-intelligence research rose to prominence.
After we spoke, Li told me she has her eyes set firmly on the longer term of AI and the hard problems that lie ahead for the sector.
Listed here are some highlights from our conversation.
Why she disagrees with a number of the AI “godfathers” about catastrophic AI risks: Other AI heavyweights, resembling Geoffrey Hinton, Yann LeCun, and Yoshua Bengio, have been jousting in public concerning the risks of AI systems and the right way to govern the technology safely. Hinton, specifically, has been vocal about his concerns that AI could pose an existential risk to humanity. Li is less convinced. “I absolutely respect that. I feel, intellectually, we should always discuss all this. But in the event you ask me as an AI leader… I feel there are other risks which are what I’d call catastrophic risks to society which are more pressing and urgent,” she says. Li highlights practical, “rubber meets the road” problems resembling misinformation, workforce disruption, bias, and privacy infringements.
Hard problems: One other major AI risk Li is worried about is the increasingly concentrated power and dominance of the tech industry on the expense of investment in science and technology research in the general public sector. “AI is so expensive—a whole bunch of hundreds of thousands of dollars for one large model, making it not possible for academia. Where does that leave science for public good? Or diverse voices beyond the shopper? America needs a moon-shot moment in AI and to significantly spend money on public-sector research and compute capabilities, including a National AI Research Resource and labs just like CERN. I firmly consider AI will help the human condition, but not and not using a coordinated effort to make sure America’s leadership in AI,” she told us.
The failings of ImageNet: ImageNet, which Li created, has been criticized for being biased and containing unsafe or harmful photos, which in turn result in biases and harmful outcomes in AI systems. Li admits the database just isn’t perfect. “It takes people to call out the imperfections of ImageNet and to call out fairness issues. Because of this we want diverse voices,” she says. “It takes a village to make technology higher.”
Copyright and data: Critics have said that the present practice of hoovering up data off the web to create data sets results in bias and copyright and privacy violations, amongst other problems. Li says the AI community must look into this. “Our collective understanding of the role data plays is so far more sophisticated now… in comparison with 2009, once we were the primary to call out the importance of knowledge,” she says, adding that the AI community needs to maintain developing and learning from past mistakes.
Suggestions for anyone wanting to get into AI: “Irrespective of where you come from, what background you may have, in the event you’re keen about this, you may have a spot [in AI],” she says. She says that folks from diverse backgrounds should try to disregard the way in which the sector is portrayed by the media, which is to say very pale and male.
Her message to young technologists? “Math is clean, but technology’s social impact is messy. Recognize that messiness, because what we’re creating has each positive and negative impact.”
One last item before you go … This week MIT Technology Review is hosting our flagship EmTech event, where we will likely be taking a look at where AI and other cutting-edge technologies go next.
Join us November 14-15, 2023, on the MIT Campus and online. Subscribers to The Algorithm get a special 30% discount on tickets! Enter the code THAGTH23EM here.
Deeper Learning
Noise-canceling headphones could let you decide and select the sounds you need to hear
Future versions of the technology could let users opt back in to certain sounds they’d prefer to hear, resembling babies crying, birds tweeting, or alarms ringing. The technology that makes it possible, called semantic hearing, allows the wearer to filter out some sounds while boosting others.
How it really works: The system, which remains to be in prototype, connects off-the-shelf noise-canceling headphones to a smartphone app. The microphones embedded in these headphones, that are used to cancel out noise, are repurposed to also detect the sounds on the earth across the wearer.
The large picture: Researchers have long tried to resolve the “cocktail party problem”—that’s, to get a pc to give attention to a single voice in a crowded room, as humans are in a position to do. This recent method represents a big step forward and demonstrates the technology’s potential. For instance, it could pave the way in which for smarter hearing aids and earphones. Read more from Rhiannon WIlliams here.
Bits and Bytes
Are we alone within the universe?
We’re getting closer than ever before to learning how common worlds like ours actually are. Latest tools, including machine learning and artificial intelligence, could help scientists look past their preconceived notions of what constitutes life. (MIT Technology Review)
Amazon and Musk have joined the AI-language-model bandwagon
ChatGPT wowed the world and kick-started Big Tech’s AI arms race around a yr ago. Latest players are still joining the race with models they hope will help them stand out. Amazon is reportedly constructing its own ChatGPT rival, code-named Olympus, which could be one in every of the largest models ever trained. Meanwhile, Elon Musk’s recent AI enterprise, xAI, debuted a pilot of its AI model, called Grok, which has been trained on data from X and is designed to reply questions “with a little bit of wit” and “a rebellious streak,” whatever meaning.
Generative AI has made Google search weird
AI-generated text is poisoning the web. Nonsense generated by language models is getting scraped into online search results, making Google search the least reliable it’s ever been for clear, accessible facts. (The Atlantic)
Could this AI pin kill the smartphone?
Silicon Valley is buzzing about an AI lapel pin created by startup Humane. The $699 device magnetically clips onto your clothes and uses a camera and sensors to record its surroundings. The thought is that folks will find a way to regulate it using their voices, and the pin will project information into people’s palms. The system uses AI models to reply questions and summarize information. The voice-controlled element makes me pause probably the most—I prefer to doomscroll discreetly in silence. (The Latest York Times)