philip lelyveld The world of entertainment technology

How Does AI ‘Think’? We Are Only Starting to Understand That

PhilNote: This is an excellent update primer.

... If you’ve gotten this far, here’s the payoff: If today’s large language models are capable of some amount of reasoning, however elementary, it could yield what could be years of rapid advances in the abilities of generative AIs.

In part, that’s because language isn’t just another medium of communication, like pictures or sound. It’s a technology humans developed for describing absolutely everything in the world we can conceive of, and how it all relates. Language gives us the ability to build models of the world, even absent any other stimuli, like vision or hearing, says Aguera y Arcas. ...

That is why a large language model can write fluently about the relationship between, say, two colors, even though it has never “seen” either of them, he says. ...

...we might soon have AI-based assistants that are completely personalized to data specific to us. Google is already attempting a first version of this—an update to its Bard generative AI allows it to search and synthesize across all of your emails, calendar items and documents, as long as they are already in Google’s system—but it’s primitive and prone to error. ...

Between the invention of the steam engine and the debut of the locomotive, more than a century elapsed. Meanwhile, a new science was born, which in turn became the midwife of countless other advancements essential to the Industrial Revolution. If the development of generative AIs conforms to this pattern at all, its near future will include transformative inventions—AIs expert in different subjects, truly personal assistants—followed by years of refinement, mad scrambles to harness and benefit from these new technologies, and possibly another sort of Industrial Revolution. But rather than a revolution predicated on energy and matter, this one will be based on the manipulation of data and insight. ...

See the full story here: https://www.wsj.com/tech/ai/how-does-ai-think-95f6381b