4 ms·

Firstly, AI is currently massively overhyped and people are overestimating the progress we are making. What we are currently doing is curve fitting and searchin

Firstly, AI is currently massively overhyped and people are overestimating the progress we are making. What we are currently doing is curve fitting and searching on steroids enabled by the ever increasing availability of computing power, what we know about general intelligence hasn't changed much in a long time, at least as far as I know. I won't rule out that we may build an AGI by just throwing enough computing power at it and simply simulating brains with enough fidelity but without actually understanding how it works, at least not at first, but I really don't see much - if any - progress towards building an AGI because we know what we have to do. Maybe the secret of general intelligence is just curve fitting and searching at massive scale and we fail to find the secret behind it because there is none, but we really don't know. That of course doesn't mean that the recent developments are not interesting or don't have useful applications, they are just not as groundbreaking as often portrayed and we probably still have a rather long way to go. With regards to big picture thinking and general problem solving ability, I don't think you are in a particularly bad spot as a software developer or hacker. Especially not if you are doing project work or frequently change jobs and regularly have to get familiar with new business domains. Also if your focus is more at the design and architecture end of the spectrum as compared to coming up with beautiful looking CSS. But I would certainly agree that philosophers have some ways of thinking that are at least to some degree unlike what you find in software developers, but I can't really pin it down. I only know it exists because I am frequently somewhat surprised by the thoughts and ideas that come up in philosophy and they seem not like things I would think of. Maybe it is just the level to which philosophers dissect things and which goes beyond what is required in software development but I really can't tell for sure.

debt 9y ago
“Firstly, AI is currently massively overhyped and people are overestimating the progress we are making.“ And I love AI! What programmer doesn’t? But I completely agree and I think it’s really difficult for people to understand why this it true!
unclebucknasty 9y ago
I tend to think that we'll continue to be stuck for a while, but that a fundamental breakthrough will usher in something approaching AGI. And, I don't think it will be along the lines of the relatively brute force current approaches. So, when it finally happens, it'll move fast.
gh02t 9y ago
I think another aspect of the current boom in AI is that it is model-less curve fitting (to use your description), and that this is both revolutionary and regressive. On the one hand machine learning algorithms let us solve fuzzy problems that we don't necessarily understand deeply, but the trade off is that those solutions don't give us any real fundamental insight into the nature of the problem the same way e.g. discovering new physics would. So we are now able to solve all these problems that we couldn't before, but in a sense these solutions are unfulfilling because they reveal little else.

debt 9y ago

“Firstly, AI is currently massively overhyped and people are overestimating the progress we are making.“ And I love AI! What programmer doesn’t? But I completely agree and I think it’s really difficult for people to understand why this it true!

unclebucknasty 9y ago

I tend to think that we'll continue to be stuck for a while, but that a fundamental breakthrough will usher in something approaching AGI. And, I don't think it will be along the lines of the relatively brute force current approaches. So, when it finally happens, it'll move fast.

gh02t 9y ago

I think another aspect of the current boom in AI is that it is model-less curve fitting (to use your description), and that this is both revolutionary and regressive. On the one hand machine learning algorithms let us solve fuzzy problems that we don't necessarily understand deeply, but the trade off is that those solutions don't give us any real fundamental insight into the nature of the problem the same way e.g. discovering new physics would. So we are now able to solve all these problems that we couldn't before, but in a sense these solutions are unfulfilling because they reveal little else.