Live data from Hacker News

The average chess players of Bletchley Park and AI research in Britain

blogs.bl.uk

41–46 of 46 posts

Re: The average chess players of Bletchley Park and AI research in Britain

#41
I remember an interview with Kasparov. He said something, I don't remember exactly... It was something like "The skills chess develops are very important for... playing chess"; as a way to say "if you're good in chess, that doesn't mean you're particularly smart or good in other areas too".

As someone who played chess competitively in my childhood and teens, chess helped me a lot about concentration, problem solving and decision taking. I also learned to win and lose and to have respect for other people due to the competition.

As a teacher in my adulthood, I was extremely impressed by knowing a high rated player that was very weak student, especially in logic.

I now agree deeply with Kasparov about the importance of the skills chess develops.

Re: The average chess players of Bletchley Park and AI research in Britain

#42

It's strange today to remember that playing chess well was seen as a great marker of AI, but today we consider it much less so. I thought Turing's Test would be a good barometer of AI, but in today's World of mountains of AI slop fooling more and more people, and ironically there being software that is better at solving CAPTCHAs than humans, I'm not so sure. Add into the mix that there are reports of people developin…

I'm not sure that there is more to it than continuous learning. If an LLM of top-tier strength could learn from its experiences even the way a junior developer could, I'm not sure I can place an upper bound on how capable it would be at software engineering. But from what I understand this will require a completely different architecture.

Re: The average chess players of Bletchley Park and AI research in Britain

#43

Earlier quoted context omitted.

I think with the Turing test, it's turned out to be a fuzzier line than expected. People are to various degrees learning LLM tells even as they improve. So what might have passed the Turing test in 2020 might not today. Similarly it seems to be a case that conversations with LLMs often start better than they end, even today - so an LLM might pass a short turing test but fail a very long one that goes into hundreds of…

We’ve clearly passed the Turing test I think. I can’t think of many ways I’d be able to detect an LLM reliably, if it was coded to just act as a person talking to me on discord.

The Turing test isn't dead. The true Turing test is a thought experiment, and it's not something that can be replicated in the real world.

Given enough time and interaction, you can still spot a person on Discord being faked by an LLM—at the very least, something will feel off. This is even more true in a formal, knowing, adversarial setting.

Re: The average chess players of Bletchley Park and AI research in Britain

#44
post #37

It's strange today to remember that playing chess well was seen as a great marker of AI, but today we consider it much less so. I thought Turing's Test would be a good barometer of AI, but in today's World of mountains of AI slop fooling more and more people, and ironically there being software that is better at solving CAPTCHAs than humans, I'm not so sure. Add into the mix that there are reports of people developin…

> I thought Turing's Test would be a good barometer of AI Depends on what you consider a "Turing's Test". Fooling unsuspecting humans is relatively easy, it has been done with relatively simple software and some trickery. LLMs can do that too of course. A more convincing "Turing's Test" would be: - You have one interrogator, and two players, one human and one computer - The interrogator, after chatting with both play…

I feel like we could code that up right now and test that. The only problem is matching human expertise.

Re: The average chess players of Bletchley Park and AI research in Britain

#45
post #34
post #32

Earlier quoted context omitted.

Robotic legs are hard, but also pretty wells solved. See Boston Dynamics.

Legs are a small subset of the problem. “Where did I see that last?” involves mapping out and classifying the environment. There’s some really impressive demos of manipulating objects on a table etc, but random clutter throughout a house is still a problem.

One interesting thing there is that we may treat a robot differently than a dog.

A dog will walk around the house, look at everything, smell everything, and touch many things. It makes things like finding slippers much easier.

I'm not sure we'd enjoy a robot doing the same, randomly wandering around the house and poking at stuff for some unspecified and uncommanded purpose.

Re: The average chess players of Bletchley Park and AI research in Britain

#46
post #37

It's strange today to remember that playing chess well was seen as a great marker of AI, but today we consider it much less so. I thought Turing's Test would be a good barometer of AI, but in today's World of mountains of AI slop fooling more and more people, and ironically there being software that is better at solving CAPTCHAs than humans, I'm not so sure. Add into the mix that there are reports of people developin…

> I thought Turing's Test would be a good barometer of AI Depends on what you consider a "Turing's Test". Fooling unsuspecting humans is relatively easy, it has been done with relatively simple software and some trickery. LLMs can do that too of course. A more convincing "Turing's Test" would be: - You have one interrogator, and two players, one human and one computer - The interrogator, after chatting with both play…

The entire premise is wrong. Humanity isn’t defined by expertise. Most average people will reach the “I don’t know” or making-it-up stage on most subjects pretty fast.

Perfect responses are more Likely indicative of a machine than a person.

Post reply on HN