Can AI be smart enough to be a threat while also being dumb enough to be a threat?
With the back and forth surrounding Sam Altman CEO of OpenAI, the spotlight is back on Artificial Intelligence (AI). Altman was fired from OpenAI only to make a triumphant return just days later. Rumour has it that Altman was ousted due to not being completely honest with OpenAI’s board about the capabilities of Q*, a top of the line research project at the firm. Some posters on the OpenAI community forum immediately started speculating that Q* was the breakthrough which will make sentient and sapient AI possible. Reality is probably more disappointing, Reuters reports that Q* is probably mainly good for making sure Chat GPT will be able to consistently get grade school math right. To say that the discussion surrounding AI is polarised between those who think it is the single most important invention since fire and those who see it as a gimmicky software built on industrialised plagiarism would be the understatement of the decade.
Much of the discussion around AI has been around its destructive potential. I find that a tad early to be worried about given how limited the current AI is, but the discussion persists. As I’ve written about before, AI programmers are likely to overestimate the potential and importance of their area of expertise and interest. Naturally the media trusts AI experts to know a lot about AI and take those experts’ overblown ideas about what AI could be capable of and regular people defer to the narratives experts in the media offer so there are few voices questioning the experts. But it is always worth questioning nonsensical claims about the total transformation of society a new Silicon Valley technology will supposedly inevitably bring.
The idea that AI will take over the world or destroy humanity is laughable. Again, OpenAI is apparently working on tech to make their top-of-the-line AI be able to solve fourth grade math problems. If humanity gets wiped out by software which doesn’t know basic division and the Pythagorean theorem, we had it coming. This leads us to what I call the ‘Idiot Savant’ problem: the notion that AI will be smart enough to be a credible threat to humanity while also not being smart enough to be reasoned with. A very convenient and highly unlikely level of intelligence, especially when it was designed by humans, who would prefer not to go extinct at the hands of their creation. In short, AI can’t be smart enough to be a threat while also being dumb enough to be a threat, one has to give.
Altman is the tech founder with a magical shortcut to solve society’s ills du jour and will remain so until people realise that large language models are useful tools which will lead to some minor efficiency increases in office work and not the imagined reordering of society. Then someone else will present the next technology to kick off a similar hype cycle, as happened with crypto before AI was cool. The idiot savant problem is an obvious issue with the ‘AI as civilizational threat’ narrative when you assess it critically. Remember that next time some big AI breakthrough is in the news.
If you liked this post you can read my last post about Ukraine here, or the rest of my writings here. It'd mean a lot to me if you recommended the blog to a friend or coworker. Come back next Monday for a new post!
I've always been interested in politics, economics, and the interplay between. The blog is a place for me to explore different ideas and concepts relating to economics or politics, be that national or international. The goal for the blog is to make you think; to provide new perspectives.
Written by Karl Johansson
Sources:
Cover photo by Igovar Igovar from Pexels, edited by Karl Johansson
Comentários