Creating “humanlike minds”
Econ Extra Credit is teaming up with “Marketplace Tech” this month to examine the artificial intelligence boom through an influential AI film: 2013’s “Her.” Subscribe here to get the whole series in your inbox.
Most of today’s artificial intelligence is what’s known as narrow AI, meaning machines that are designed to execute a single, specific task.
Think of “Deep Blue,” an AI-driven chess program run on an IBM supercomputer that successfully beat world champion Garry Kasparov in a six-game match in 1997. While narrow AI can copy humans’ abilities to problem-solve and learn, these machines must still be programmed by humans.
That’s far more simplistic than the AI we’re used to seeing in science fiction, especially in movies. The superintelligent humanoids of “Westworld,” the malevolent supercomputer in “2001: A Space Odyssey” and the emotionally attuned operating system in “Her” are all akin to artificial general intelligence, or AGI, in which machines are capable of everything humans are, or even more.
So where’s the tipping point between narrow AI and AGI? It’s hotly debated, in part because there’s not one universal definition of AGI.
“We still can’t come up with a consensus on what it means to be human, so we’re not going to come up with consistent answers on whether or not what we have is AGI,” said John Licato, a professor of computer science at the University of South Florida.
Alan Turing, an early pioneer of artificial intelligence, developed a simple test for AGI. The Turing test is meant to determine whether computers can think and talk like a human such that a real human could not tell the difference. But it’s not the be-all, end-all. Apple co-founder Steve Wozniak theorized a machine could exhibit AGI abilities if it could enter his home and make a cup of coffee, with little or no explicit instructions.
“I think, though, that a lot of people are able to say that the reasoning used by current language models is crossing a line, and it’s reasonable to say that yes, it actually is closer to AGI than it isn’t,” Licato told Meghan McCarty Carino on “Marketplace Tech.”
So how close are we to AGI now? Researchers at Microsoft published a paper that refers to “sparks of AGI” in GPT-4, and several Google leaders who were interviewed on “60 Minutes” referred to “emergent properties” in one of their language models, including translating languages they said the model had not been explicitly trained on. But these claims have been heavily criticized by some AI researchers, including Google’s own former co-lead of AI ethics, Margaret Mitchell.
Even though many researchers would still say that AGI doesn’t exist, Licato estimates that “Her”-level tech could be only a few years away.
“If I had to bet, I would say under five years,” he said.
“The underlying technology that makes state-of-the-art language models possible is something that we’ve had since probably the 1980s,” Licato said. “It’s just that we never had the available data in the sizes that we have now, and we didn’t have the compute power that we did to actually process it.”
“So, it might be that we already have the algorithmic know-how that we need to reach AGI,” Licato added. “It’s just that we’re waiting to have enough computational power to hit it. It could be that we’re already on that path.”
“Her” is available to stream on Criterion Channel with a subscription. You can also rent or buy it on many platforms, including Prime Video, Apple TV, Vudu and YouTube.
After you watch, send us your thoughts and questions at extracredit@marketplace.org or reply to this email!
There’s a lot happening in the world. Through it all, Marketplace is here for you.
You rely on Marketplace to break down the world’s events and tell you how it affects you in a fact-based, approachable way. We rely on your financial support to keep making that possible.
Your donation today powers the independent journalism that you rely on. For just $5/month, you can help sustain Marketplace so we can keep reporting on the things that matter to you.
