It was leaked that you had an exchange with Elon Musk regarding the risks posed by AI. [Ed note: Musk said he had told the Chinese government that AI might eventually be able to overtake it, and Raji responded by questioning the safety of today’s driverless cars, like the autopilot feature in a Tesla.] Can you tell me more about that?
You know, it wasn’t just Elon. That was the one that got out. There was another CEO that was talking about curing cancer with AI, saying we have to make sure that it’s Americans that do that, and just narratives like that.
But first of all, we have medical AI technology that is hurting people and not working well for Black and brown patients. It’s disproportionately underprioritizing them in terms of getting a bed at a hospital; it’s disproportionately misdiagnosing them, and misinterpreting lab tests for them.
I also hope that one day AI will lead to cancer cures, but we need to understand the limitations of the systems that we have today.
What was it that you really wanted to achieve in the forum, and do you think you had the chance to do that?
I think we all had substantial opportunities to say what we needed to say. In terms of whether we were all equally heard or equally understood, I think that’s something that I’m still processing.
My main position coming in was to debunk a lot of the myths that were coming out of these companies around how well these systems are working, especially on marginalized folks. And then also to debunk some of the myths around solving bias and fairness.
Bias concerns and explainability concerns are just really difficult technical and social challenges. I came in being like, I don’t want people to underestimate the challenge.
So did I get that across? I’m not sure, because the senators loved saying that AI is gonna cure cancer.
It’s so easy to get caught up in the marketing terms and the sci-fi narratives and completely ignore what’s happening on the ground. I’m coming back from all of this more committed than ever to articulating and demonstrating the reality, because it just seems like there is this huge gap of knowledge between what’s actually happening and the stories that these senators are hearing from these companies.
What else I’m reading
- I just loved this story from Jessica Bennett at the New York Times about what it’s like to be a teen girl with a cell phone today. Bennett kept in touch with three 13-year-olds over the course of a year to learn about the ins and outs of their digital lives. Highly recommend!
- This social reflection on privacy by Charlie Warzel in the Atlantic has stuck with me for a few days. The story gets at the overwhelming questions we—certainly I—have about what we can do to preserve our privacy online.
- The United Nations General Assembly convened in New York this past week, and one big topic of discussion was, of course, AI. Will Henshall at Time did a deep dive into what we might expect from the body on AI regulation.
What I learned this week
A Disney director tried to use AI to create a soundtrack reminiscent of the work of symphonist Hans Zimmer—and came up disappointed. Gareth Edwards, director of Rogue One: A Star Wars Story, told my colleague Melissa Heikkilä that he was hoping to use AI to create a soundtrack for his forthcoming movie about … AI, of course! Well, the soundtrack fell flat, and Edwards even shared it with the famous composer, who he says found it amusing.
Melissa wrote, “Edwards said AI systems lack a fundamentally crucial skill for creating good art: taste. They still don’t understand what humans deem good or bad.”
In the end, the real Zimmer wrote the melodies for Edwards’s upcoming movie, The Creator.
The Download: COP28 controversy and the future of families
The United Arab Emirates is one of the world’s largest oil producers. It’s also the site of this year’s UN COP28 climate summit, which kicks off later this week in Dubai.
It’s a controversial host, but the truth is that there’s massive potential for oil and gas companies to help address climate change, both by cleaning up their operations and by investing their considerable wealth and expertise into new technologies.
The problem is that these companies also have a vested interest in preserving the status quo. If they want to be part of a net-zero future, something will need to change—and soon. Read the full story.
How reproductive technology can reverse population decline
Birth rates have been plummeting in wealthy countries, well below the “replacement” rate. Even in China, a dramatic downturn in the number of babies has officials scrambling, as its population growth turns negative.
So, what’s behind the baby bust and can new reproductive technology reverse the trend? MIT Technology Review is hosting a subscriber-only Roundtables discussion on how innovations from the lab could affect the future of families at 11am ET this morning, featuring Antonio Regalado, our biotechnology editor, and entrepreneur Martín Varsavsky, founder of fertility clinic Prelude Fertility. Don’t miss out—make sure you register now.
Unpacking the hype around OpenAI’s rumored new Q* model
While we still don’t know all the details, there have been reports that researchers at OpenAI had made a “breakthrough” in AI that had alarmed staff members. Reuters and The Information both report that researchers had come up with a new way to make powerful AI systems and had created a new model, called Q* (pronounced Q star), that was able to perform grade-school-level math. According to the people who spoke to Reuters, some at OpenAI believe this could be a milestone in the company’s quest to build artificial general intelligence, a much-hyped concept referring to an AI system that is smarter than humans. The company declined to comment on Q*.
Social media is full of speculation and excessive hype, so I called some experts to find out how big a deal any breakthrough in math and AI would really be.
Researchers have for years tried to get AI models to solve math problems. Language models like ChatGPT and GPT-4 can do some math, but not very well or reliably. We currently don’t have the algorithms or even the right architectures to be able to solve math problems reliably using AI, says Wenda Li, an AI lecturer at the University of Edinburgh. Deep learning and transformers (a kind of neural network), which is what language models use, are excellent at recognizing patterns, but that alone is likely not enough, Li adds.
Math is a benchmark for reasoning, Li says. A machine that is able to reason about mathematics, could, in theory, be able to learn to do other tasks that build on existing information, such as writing computer code or drawing conclusions from a news article. Math is a particularly hard challenge because it requires AI models to have the capacity to reason and to really understand what they are dealing with.
A generative AI system that could reliably do math would need to have a really firm grasp on concrete definitions of particular concepts that can get very abstract. A lot of math problems also require some level of planning over multiple steps, says Katie Collins, a PhD researcher at the University of Cambridge, who specializes in math and AI. Indeed, Yann LeCun, chief AI scientist at Meta, posted on X and LinkedIn over the weekend that he thinks Q* is likely to be “OpenAI attempts at planning.”
People who worry about whether AI poses an existential risk to humans, one of OpenAI’s founding concerns, fear that such capabilities might lead to rogue AI. Safety concerns might arise if such AI systems are allowed to set their own goals and start to interface with a real physical or digital world in some ways, says Collins.
But while math capability might take us a step closer to more powerful AI systems, solving these sorts of math problems doesn’t signal the birth of a superintelligence.
“I don’t think it immediately gets us to AGI or scary situations,” says Collins. It’s also very important to underline what kind of math problems AI is solving, she adds.
The Download: unpacking OpenAI Q* hype, and X’s financial woes
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Unpacking the hype around OpenAI’s rumored new Q* model
Ever since last week’s dramatic events at OpenAI, the rumor mill has been in overdrive about why the company’s board tried to oust CEO Sam Altman.
While we still don’t know all the details, there have been reports that researchers at OpenAI had made a “breakthrough” in AI that alarmed staff members. The claim is that they came up with a new way to make powerful AI systems and had created a new model, called Q* (pronounced Q star), that was able to perform grade-school level math.
Some at OpenAI reportedly believe this could be a breakthrough in the company’s quest to build artificial general intelligence, a much-hyped concept of an AI system that is smarter than humans.
So what’s actually going on? And why is grade-school math such a big deal? Our senior AI reporter Melissa Heikkilä called some experts to find out how big of a deal any such breakthrough would really be. Here’s what they had to say.
This story is from The Algorithm, our weekly newsletter giving you the inside track on all things AI. Sign up to receive it in your inbox every Monday.
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 X is hemorrhaging millions in advertising revenue
Internal documents show the company is in an even worse position than previously thought. (NYT $)
+ Misinformation ‘super-spreaders’ on X are reportedly eligible for payouts from its ad revenue sharing program. (The Verge)
+ It’s not just you: tech billionaires really are becoming more unbearable. (The Guardian)
2 The brakes seem to now be off on AI development
With Sam Altman’s return to OpenAI, the ‘accelerationists’ have come out on top. (WSJ $)
+ Inside the mind of OpenAI’s chief scientist, Ilya Sutskever. (MIT Technology Review)
3 How Norway got heat pumps into two-thirds of its households
Mostly by making it the cheaper choice for people. (The Guardian)
+ Everything you need to know about the wild world of heat pumps. (MIT Technology Review)
4 How your social media feeds shape how you see the Israel-Gaza war
Masses of content are being pumped out, rarely with any nuance or historical understanding. (BBC)
+ China tried to keep kids off social media. Now the elderly are hooked. (Wired $)
5 US regulators have surprisingly little scope to enforce Amazon’s safety rules
As demonstrated by the measly $7,000 fine issued by Indiana after a worker was killed by warehouse machinery. (WP $)
6 How Ukraine is using advanced technologies on the battlefield
The Pentagon is using the conflict as a testbed for some of the 800-odd AI-based projects it has in progress. (AP $)
+ Why business is booming for military AI startups. (MIT Technology Review)
7 Shein is trying to overhaul its image, with limited success
Its products seem too cheap to be ethically sourced—and it doesn’t take kindly to people pointing that out. (The Verge)
+ Why my bittersweet relationship with Shein had to end. (MIT Technology Review)
8 Every app can be a dating app now
As people turn their backs on the traditional apps, they’re finding love in places like Yelp, Duolingo and Strava. (WSJ $)
+ Job sharing apps are also becoming more popular. (BBC)
9 People can’t get enough of work livestreams on TikTok
It’s mostly about the weirdly hypnotic quality of watching people doing tasks like manicures or frying eggs. (The Atlantic $)
10 A handy guide to time travel in the movies
Whether you prioritize scientific accuracy or entertainment value, this chart has got you covered. (Ars Technica)
Quote of the day
“It’s in the AI industry’s interest to make people think that only the big players can do this—but it’s not true.”
—Ed Newton-Rex, who just resigned as VP of audio at Stability.AI, says the idea that generative AI models can only be built by scraping artists’ work is a myth in an interview with The Next Web.
The big story
The YouTube baker fighting back against deadly “craft hacks”
Ann Reardon is probably the last person you’d expect to be banned from YouTube. A former Australian youth worker and a mother of three, she’s been teaching millions of subscribers how to bake since 2011.
However, more recently, Reardon has been using her platform to warn people about dangerous new “craft hacks” that are sweeping YouTube, such as poaching eggs in a microwave, bleaching strawberries, and using a Coke can and a flame to pop popcorn.
Reardon was banned because she got caught up in YouTube’s messy moderation policies. In doing so, she exposed a failing in the system: How can a warning about harmful hacks be deemed dangerous when the hack videos themselves are not? Read the full story.
We can still have nice things
+ London’s future skyline is looking increasingly like New York’s.
+ Whovians will never agree on who has the honor of being the best Doctor.
+ How to get into mixing music like a pro.
+ This Japanese sea worm has a neat trick up its sleeve—splitting itself in two in the quest for love.
+ Did you know there’s a mysterious tunnel under Seoul?