Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Past Performance is Not Indicative of Future Results (2020)

locusmag.com

Post reply on HN
Print view View original

151–160 of 285 posts

Re: Past Performance is Not Indicative of Future Results (2020)

#151

Post by skohan »

version_five wrote: post #149
skohan wrote: post #109

Earlier quoted context omitted.

Yeah I agree - during undergrad, I spent a few years studying neuroscience, and I was very let down by my first ML/AI course. Compared to what I had learned about the brain, what we called an "ANN" just seemed like such a silly toy. The more you learn about neurobiology, the more apparent it is that there are so many levels of computation going on - everything from dendritic structure, to cellular metabolism, to epig…

So you took an undergrad ML course and you're using this as the basis for your conclusions about how ML can scale? You understand modern neural networks as large matrix operations and then attack that idea leading to intelligence as a joke? I also find it improbable that intelligence will emerge from modern ML without some major leap. But you have added nothing to the discussion, beyond some impressions from undergra…

I'm sorry, I don't mean to insult or offend anyone. I'm just recounting my observations based on my understanding of the subject - and that is really not to disparage the amazing work that's being done, but rather to highlight the scale of the problem you have to solve when you're talking about creating something similar to human intelligence. It's entirely possible I'm wrong about this, and I would love to be proven so.

Do you disagree substantively with anything I have said, or do you just think I could have phrased it better?

Re: Past Performance is Not Indicative of Future Results (2020)

#152

Post by dane-pgp »

okareaman wrote: post #94
Trasmatta wrote: post #86

Earlier quoted context omitted.

If our brains are receivers to a field of consciousness, why would it be impossible to replicate one of those receivers with a machine? You also seem to have just kicked the can down the road. "Consciousness arises from a field somehow, and the brain acts as a receiver somehow. The somehow is not explained."

I didn't say I knew how. I said I believe materialism is a dead end, by which I mean I doubt the consciousness arises out of atoms configured as neurons. How those neurons receive a conscious field seems a more productive line of inquiry, but for some reason people resist this idea. Not sure why.

By studying the atoms configured as neurons, we've managed to develop machines that can learn to play board games and Atari games better than humans, and can write prose and poetry at a convincingly human level. Those skills may not require consciousness, but it's not clear that these machines would be more useful if they could "receive a conscious field".

Do you think that animals receive a conscious field? Could we create an accurate representation of a mouse's brain just from modelling its neurons? If a mouse brain can't receive a conscious field, but a human brain can, then what relevant physiological differences are there between the two, other than size?

Re: Past Performance is Not Indicative of Future Results (2020)

#153

Post by dtech »

gonehome wrote: post #135
skohan wrote: post #109

Earlier quoted context omitted.

Yeah I agree - during undergrad, I spent a few years studying neuroscience, and I was very let down by my first ML/AI course. Compared to what I had learned about the brain, what we called an "ANN" just seemed like such a silly toy. The more you learn about neurobiology, the more apparent it is that there are so many levels of computation going on - everything from dendritic structure, to cellular metabolism, to epig…

I'm not sure I buy that - biology is often messier because of nature related constraints, it gets optimized for other things (energy, head size, etc.) The way a plane flies is quite different than the way a bird flies in complexity - they share an underlying mechanism, but planes don't need to flap wings. It's possible that scaling up does lead to generality and we've seen hints of that. - https://deepmind.com/blog/a…

> It’s hard to predict timelines for this kind of thing, and people are notoriously bad at it. Few would have predicted the results we’re seeing today in 2010. What would you expect to see in the years leading up to AGI? Does what we’re seeing look like failure?

Few have predicted a reasonably-capable text-writing engine or automatic video face replacement, but many have predicted self-driving cars would have been readily available to consumers by now and semi-intelligent helper-robots being around.

Just because unforeseen advancements have been made, does not mean that foreseen advancements come true.

Re: Past Performance is Not Indicative of Future Results (2020)

#154

Post by yarg »

Past performance is not indictative of future results across distinct domains.

Within a single problem space (or sub-space) past performance can generalise quite well.

There's a problem with scaling solutions and expecting performance to continue to increase in a continuous exponential manner: growth that we perceive as exponential is often only on a long-life S-Curve.

We've seen this in silicon, where what appears to the layman to have been exponential growth has in fact been a sequence of more limited growth spurts bound by the physical limits of scaling within whatever model of design was active at the time.

The question of where the bounds to the problem domains are, and when new ideas or paradigms are required is much more difficult in AI than it has been in microprocessors.

It's easy enough to formulate the question "how small can this be before the changes in physical characteristics at scale prevent it from working?", if rather more difficult to answer.

AI is so damned steeped in the vagaries of the unknown that I can't even think of the question.

Re: Past Performance is Not Indicative of Future Results (2020)

#155

Post by version_five »

skohan wrote: post #151
version_five wrote: post #149

Earlier quoted context omitted.

So you took an undergrad ML course and you're using this as the basis for your conclusions about how ML can scale? You understand modern neural networks as large matrix operations and then attack that idea leading to intelligence as a joke? I also find it improbable that intelligence will emerge from modern ML without some major leap. But you have added nothing to the discussion, beyond some impressions from undergra…

I'm sorry, I don't mean to insult or offend anyone. I'm just recounting my observations based on my understanding of the subject - and that is really not to disparage the amazing work that's being done, but rather to highlight the scale of the problem you have to solve when you're talking about creating something similar to human intelligence. It's entirely possible I'm wrong about this, and I would love to be proven…

Thanks for your reply. I suppose a quick way to summarize my criticism is that it reads to me like you've dismissed the strengths of ML on technical grounds, while you imply you don't have any real technical experience in the field. You make a superficial comparison between the compexity of biology and ML, without providing any real insight, just saying one has lots going on and the other is matrix multiplication.

If your conclusion is that current gradient based methods probably won't scale up to AGI, you're probably right. But if you want to get involved in the discussion of why this is true, what ML actually can and can't do, etc. I would encourage you to learn more about the subject and the current research areas, and draw on that for your discussion points.

Otherwise, it comes across as "I once saw a podcast that said..." type stuff that is hard to take seriously.

No doubt I come across as condescending, please take what I say with the usual weight you'd assign to the views of a random guy on the internet :)

Re: Past Performance is Not Indicative of Future Results (2020)

#157

Post by dvt »

What a confused and muddled post, trying to touch on psychology, philosophy, and mathematics, and missing the mark on basically all three. I'm quite bearish on AI/ML, but calling it a "parlor trick" is like calling modern computers a parlor trick. I mean, at the end of the day, they're just very fast abacuses, right? Let's face it: what ML has brought to the forefront -- from self-landing airplanes to self-driving cars, to AI-assisted diagnoses -- is pretty impressive. If you insist on being reductive, sure, I guess it's "merely" statistics.

Bringing up quantitative vs qualitative analysis is just silly, since science has had this problem way before AI. Hume famously described it as the is/ought problem†. And that was a few hundred years ago.

Finally, dropping the mic with "I don't think we're anywhere close to consciousness" is just bizarre. I don't think that any serious academic working in AI/ML has made any arguments that claim machine learning models are "conscious." And Strong AI will probably remain unattainable for a very long time (I'd argue forever). This is not a particularly controversial position.

† Okay, it's not the same thing, but closely related. I suppose the fact–value distinction might be a bit closer.

Re: Past Performance is Not Indicative of Future Results (2020)

#158

Post by flyinglizard »

dvt wrote: post #157

What a confused and muddled post, trying to touch on psychology, philosophy, and mathematics, and missing the mark on basically all three. I'm quite bearish on AI/ML, but calling it a "parlor trick" is like calling modern computers a parlor trick. I mean, at the end of the day, they're just very fast abacuses, right? Let's face it: what ML has brought to the forefront -- from self-landing airplanes to self-driving ca…

> what ML has brought to the forefront -- from self-landing airplanes to self-landing cars

I am not aware of any ML in flight controls. Being black box and probabilistic by nature, these things won’t get past industry standards and regulations (at least for a while).

Re: Past Performance is Not Indicative of Future Results (2020)

#159

Post by dvt »

flyinglizard wrote: post #158
dvt wrote: post #157

What a confused and muddled post, trying to touch on psychology, philosophy, and mathematics, and missing the mark on basically all three. I'm quite bearish on AI/ML, but calling it a "parlor trick" is like calling modern computers a parlor trick. I mean, at the end of the day, they're just very fast abacuses, right? Let's face it: what ML has brought to the forefront -- from self-landing airplanes to self-driving ca…

> what ML has brought to the forefront -- from self-landing airplanes to self-landing cars I am not aware of any ML in flight controls. Being black box and probabilistic by nature, these things won’t get past industry standards and regulations (at least for a while).

> I am not aware of any ML in flight controls. Being black box and probabilistic by nature, these things won’t get past industry standards and regulations (at least for a while).

(Hah, I accidentally wrote "self-landing cars," fixed). But yeah, I guess I was thinking more of drones, I'm not exactly sure what ML (if any) is in the guts of a commercial or military airplane.

Post reply on HN