7 ms·

"In the first place, I’m skeptical. I don’t believe we have yet crossed the threshold where machines learn to solve interesting computational problems on their

"In the first place, I’m skeptical. I don’t believe we have yet crossed the threshold where machines learn to solve interesting computational problems on their own. I don’t believe we’re even close to that point, or necessarily heading in the right direction." I feel like this is pretty hard to believe, to be honest. How on Earth, given all that's happened last year, could we possibly not at least be "heading in the right direction" the way things are?

starbugs 3y ago
The author qualifies this statement towards the end of the article where he also provides a short review of a few studies: > My personal view is this: If you view the current generation of LLMs as pioneering efforts in a long-term research program, then the results are encouraging. But if you consider this technology as an immediate replacement for hand-coded software, it’s quite hopeless. We’re nowhere near the necessary level of reliability. I think especially the overview of the studies provided in the article are valuable. You can draw your own conclusions from the results.
gumby 3y ago
The automobile is faster than a human but is not a path towards machines that can transport things long distances including climbing hills or stairs or fording rivers. Generative transformers only address part of the problem space, and perhaps I should have written “appear to address”. It is unlikely that they are much of a step along the path — perhaps, to use a car analogy, they are a crankshaft. Perhaps a better analogy is the analogue computers of the 40s-60s and the later digital/analogue hybrids. Computationally interesting for the problems of the age but ultimately a dead end.
pawelmurias 3y ago
> How on Earth, given all that's happened last year, could we possibly not at > least be "heading in the right direction" the way things are? The LLM models that are all the hype are superb at things like copy writing and the more traditional humanities but struggle at logic. With a year of reasearch we can reach human parity in things like psychotherapy but for something that requires logical reasoning rather then repeating bits of text it's not that clear.
rperez333 3y ago
They are already doing it: most humans are not capable of writing a program to drive a car, be an interesting talking companion, or even remove noise from audio. We still need humans to kickstart the learning process, that’s true, but even them are not the programmers we are used too.
rramadass 3y ago
You might want to look into the works of Melanie Mitchell for some pointers; IMO she gives a balanced view with both feet planted firmly on the ground.

starbugs 3y ago

The author qualifies this statement towards the end of the article where he also provides a short review of a few studies: > My personal view is this: If you view the current generation of LLMs as pioneering efforts in a long-term research program, then the results are encouraging. But if you consider this technology as an immediate replacement for hand-coded software, it’s quite hopeless. We’re nowhere near the necessary level of reliability. I think especially the overview of the studies provided in the article are valuable. You can draw your own conclusions from the results.

gumby 3y ago

The automobile is faster than a human but is not a path towards machines that can transport things long distances including climbing hills or stairs or fording rivers. Generative transformers only address part of the problem space, and perhaps I should have written “appear to address”. It is unlikely that they are much of a step along the path — perhaps, to use a car analogy, they are a crankshaft. Perhaps a better analogy is the analogue computers of the 40s-60s and the later digital/analogue hybrids. Computationally interesting for the problems of the age but ultimately a dead end.

pawelmurias 3y ago

> How on Earth, given all that's happened last year, could we possibly not at > least be "heading in the right direction" the way things are? The LLM models that are all the hype are superb at things like copy writing and the more traditional humanities but struggle at logic. With a year of reasearch we can reach human parity in things like psychotherapy but for something that requires logical reasoning rather then repeating bits of text it's not that clear.

rperez333 3y ago

They are already doing it: most humans are not capable of writing a program to drive a car, be an interesting talking companion, or even remove noise from audio. We still need humans to kickstart the learning process, that’s true, but even them are not the programmers we are used too.

rramadass 3y ago

You might want to look into the works of Melanie Mitchell for some pointers; IMO she gives a balanced view with both feet planted firmly on the ground.