Coming Soon to a Computer Near You: Artificial ‘Superintelligence’

  

Artificial intelligence, so far, has not been all that intelligent. No AI program released to date can create anything; they just re-hash and re-arrange data they already have access to. They can’t come up with anything new on their own; no AI program is sentient or self-aware.

Advertisement

But one tech company CEO thinks that may be changing, and soon.

Open AI CEO Sam Altman says the world could be just “a few thousand days” from creating an artificial “superintelligence.”

Altman made the assertion in a personal blog post on Monday, declaring that the world is now at the “dawn of the Intelligence Age.”

“This may turn out to be the most consequential fact about all of history so far. It is possible that we will have superintelligence in a few thousand days (!); it may take longer, but I’m confident we’ll get there,” Altman wrote.

Current AIs, such as ChatGPT, are large language models that analyze troves of data to provide responses and reactions. A “superintelligent” AI, however, would have greater cognitive abilities than any human, as well as the ability to iterate on and improve itself. 

Color me skeptical.

The difference between what we are thinking of as artificial intelligence these days and what Sam Altman is proposing is so great as to be a difference not of degree but of kind. Current AI is artificial but not all that intelligent; it cannot create anything, and it can only “learn” through algorithms that are designed and created by humans, and that “learning” is just responding to prompts by rearranging existing data. When even a few young Marines can bollix up an AI program, you know that the state of the art has a long way to go.

Advertisement

See Related: AI May Be Learning to Deceive Humans – but AI Isn’t Counting on Marines 

Hey, How About No: AI Company Co-Founder Says ‘We’re Going to Have Supervision’

Granted I’m not a programmer, but I am a biologist, by education. It seems unlikely that we will ever be able to build a sentient AI, and very likely not even a “superintelligent” AI that can create, that can self-analyze, and self-improve, in a sense, to evolve. 

Here’s the thing about natural – human – intelligence. It’s chaotic, and complex beyond our capacity to truly understand. We don’t even know what consciousness and self-awareness are, what causes them, or what structures in the brain are responsible for them in any but the most superficial sense. We know humans are self-aware, but there is a spectrum in nature of self-awareness; an amoeba is not self-aware, nor is an earthworm. A chimp is self-aware, but not in the same way as humans are. And the best computer models we have created are still laughably crude compared to biological systems.

No machine, no program, has ever been self-aware; we can’t even reproduce the level of self-awareness of a dog or cat, much less a “super-intelligence” that can improve itself. And it’s not at all clear that we should try.

Advertisement

The problem is this: The AI genie is out of the bottle, and there’s no putting it back. If it’s possible to do this, to build a self-improving AI capable of creating, then someone, somewhere, will do this. If not an American company, then who’s to say it might not happen in Russia, China – or North Korea?

This seems relevant.