Gary Marcus meets me outside the post office of Vancouver’s Granville Island wearing neon-coral sneakers and a blue Arc’teryx jacket. I’m on the town for a family thing, and Marcus has lived in the town since 2018, after 20 years in Recent York City. “I just find it to be paradise,” he tells me, as I join him on his each day walk around Granville Island and nearby Kitsilano Beach. We’ve agreed to walk and speak about—what else—the present state of AI. 

“I’m depressed about it,” he tells me. “Once I went into this field, it was not in order that we could have a large turnover of wealth from artists to big corporations.” I take an enormous sip of my black dark-roast coffee. Off we go. 

Marcus, a professor emeritus at NYU, is a distinguished AI researcher and cognitive scientist who has positioned himself as a vocal critic of deep learning and AI. He’s a divisive figure. You would possibly recognize him from the spicy feuds on X with AI heavyweights resembling Yann LeCun and Geoffrey Hinton. (“All attempts to socialize me have failed,” he jokes.) It’s on walks like this that Marcus often does most of his tweeting.

This week has been an enormous news week in AI. Google DeepMind launched the following generation of its powerful artificial-intelligence model Gemini, which has an enhanced ability to work with large amounts of video, text, and pictures. And OpenAI has built a striking recent generative video model called Sora that may take a brief text description and switch it into an in depth, high-definition film clip as much as a minute long. AI video generation has been around for some time, but Sora seems to have upped the ante. My X timeline has been flooded with stunning clips people have generated using the software. OpenAI claims that its results suggest that scaling video generation models like Sora “is a promising path towards constructing general purpose simulators of the physical world.” You possibly can read more about Sora from Will Douglas Heaven here. 

But—surprise—Marcus isn’t impressed. “Should you take a look at [the videos] for a second, you’re like, ‘Wow, that’s amazing.’ But if you happen to take a look at them fastidiously, [the AI system] still doesn’t really understand common sense,” he says. In some videos, the physics are clearly off, and animals and other people spontaneously appear and disappear, or things fly backwards, for instance. 

For Marcus, generative video is yet one more example of the exploitative business model of tech firms. Many artists and writers and even thehave sued AI firms, claiming that their practice of indiscriminately scraping the web for data to coach their models violates their mental property. Copyright issues are top of Marcus’s mind. He managed to get popular AI image generators to generate scenes from Marvel movies or famous characters resembling the Minions, Sonic the Hedgehog, and Darth Vader. He has began lobbying for clearer rules on what goes into AI models.  

“Video generation mustn’t be done with copyrighted materials taken without consent, in systems which might be opaque, where we will’t understand what’s happening,” he says. “It shouldn’t be a legal thing. It’s definitely not an ethical thing.” 

We stop at a scenic spot. It’s an exquisite route, with views of the town, the mountains, and the beach. A speckle of sun hits the height of a mountain just across the bay. We couldn’t be further away from Silicon Valley, the epicenter of today’s AI boom. “​​I’m not a non secular person, but these sorts of tableaux … just proceed to blow my mind,” Marcus says. 

But despite the tranquility of the environment, it’s on walks like this that Marcus often uses X to rail against the facility structures of Silicon Valley. Right away, he says, he identifies as an activist. 

Once I ask him what motivates him, he replies without missing a beat: “The people who find themselves running AI don’t really care that much about what you may call responsible AI, and that the results for society could also be severe.” 

Late last yr he wrote a book, called , which is coming out this fall. It’s his manifesto on how AI needs to be regulated, but additionally a call to motion. “We want to get the general public involved within the struggle to attempt to get the AI firms to behave responsibly,” he says. 

There are a bunch of various things people can do, starting from boycotting a number of the software until people clean up their act to selecting electoral candidates around their tech policies, he says. 

Motion and AI policy are needed urgently, he argues, because we’re in a really narrow window during which we will sort things in AI. The chance is that we make the identical mistakes regulators made with social media firms. 

“What we saw with social media is just going to be like an appetizer in comparison with what’s going to occur,” he says. 

Around 12 000 steps later, we’re back at Granville Island’s Public Market. I’m ravenous, so Marcus shows me a spot that serves good bagels. We each get the lox with cream cheese and eat it outside within the sun before parting ways.  

Later that day, Marcus would send out a flurry of tweets about Sora, having seen enough evidence to call it: “Sora is unbelievable, but it surely is akin to morphing and splicing, slightly than a path to the physical reasoning we would want for AGI,” he wrote. “We’ll see more systemic glitches as more people have access. Many might be hard to treatment.” 

Don’t say he didn’t warn you. 

_______________________________________

DEEPER LEARNING

A brand new satellite will use Google’s AI to map methane leaks from space

A methane-measuring satellite will launch next month that goals to make use of Google’s AI to quantify, map, and reduce leaks. The mission is an element of a collaboration with the nonprofit Environmental Defense Fund, and the result, they are saying, might be probably the most detailed portrait yet of methane emissions. It should help to discover where the worst spots are and who’s responsible.

Putting methane on the map: With methane accountable for roughly a 3rd of the warming brought on by greenhouse gases, regulators within the US and elsewhere are pushing for stronger rules to curb the leaks that spring from oil and gas plants. MethaneSAT will measure the plumes of methane that billow invisibly from oil and gas operations across the globe, and Google and EDF will then map those leaks to be used by researchers, regulators, and the general public. Read more from our recent AI reporter James O’Donnell. James will cover the intersection between AI and hardware, resembling robotics and chips. 

_____________________________________________________________________

BITS AND BYTES

Is AI going to alter how we define videos? 

Systems like OpenAI’s Sora don’t make recordings. They render ideas. Does it matter that they’re not real? (Recent Yorker)

Early adopters of Microsoft’s AI bot are wondering if it’s definitely worth the money

Testers have had it of their hands for six months now, and the outcomes are mixed, to say the least. (WSJ)

The White House will spend $1.5 billion on a brand new chip factory

The huge grant, a part of the CHIPS and Science Act, will help the US establish a homegrown supply for a number of the most important components of recent life. (WP)

AI hype has echoes of the telecom boom and bust

When a chief executive asks for trillions, not billions, when raising funds you recognize a sector be getting a bit too hot. (FT)