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

This AI Boom Will Also Bust

overcomingbias.com

Post reply on HN
Print view View original

261–270 of 320 posts

Re: This AI Boom Will Also Bust

#261

Post by AndrewKemendo »

notahacker wrote: post #195
jbrambleDC wrote: post #146

Earlier quoted context omitted.

another point is that Linear Regression IS Machine/statistical Learning. Sure its been around for more than 100 years before computation, but regression algorithms are learning algorithms. Arguing for more linear regression to solve a firms problems, is equivalent to arguing for machine learning. Now, if instead he wanted to argue that the vast majority of a businesses prediction problems can be solved by simple algo…

If we're classing linear regression as machine learning and agreeing it's a representative example of the type of simple algorithm that's most likely to benefit firms, I think it probably helps his point rather than harming it. It's a technique that's been around for ages, is far from arcane knowledge and every business has had the computing capability to run useful linear regressions on various not-particularly-huge…

This overshoots the target. It's like saying that we use algebra and therefore =/= AI.

LR and general regression schemes are captured in supervised learning methods. So yes, the systems use linear regression as a fundamental attribute but build on them significantly.

Re: This AI Boom Will Also Bust

#262

Post by randcraw »

mwfunk wrote: post #215

I'm very curious to what degree there even is an AI boom right now, vs. AI and machine learning going through a phase as the buzzwords du jour used in corporate PR. People have doing all sorts of fascinating things with machine learning for decades, and (for example) Google has been arguably an AI-focused company from day one. In the tech press recently, I keep hearing how every huge tech company needs to have some s…

I work at a large pharma, and from what I've seen in this business and anywhere else where data is important (like R&D), deep learning is perceived to be revolutionary.

I think most of the rise in DL's corporate mindshare is the potential it has for disruption of the status quo. Business folks hate game changing tech. It destabilizes the priorities and practices they understand, and dishevels the grand hierarchy of corporate command and control they know and love.

If DL really does reinvent any significant fraction of their part of the business, they know 1) they won't know what to do in response, and 2) their daily competition is going to get even more challenging until a new status quo emerges. And both of which may cause them to lose their tenuous position/status in the company hierarchy, much less, lose their very job.

Re: This AI Boom Will Also Bust

#263

Post by jonathankoren »

ffwd wrote: post #222
jonathankoren wrote: post #210

Earlier quoted context omitted.

This is literally a philosophical problem. It's called ontology. And no amount of data solves this problem, because ultimately it's a labeling problem, and the border between things is ill defined, and additional data doesn't help resolve labeling ambiguity, if anything it finds out just how ill defined the world actually is. Think about it. Let's say you had a problem which was find the black squares. So you collect…

> if anything it finds out just how ill defined the world actually is. Yeah I hear this but it seems only half-true to me. While for most intents and purposes the world is ill-defined, in another sense the world itself is "100% signal" and no noise. If we "zoom out" and take a grand view, imagining that we have a supercomputer and a huge database, and the algorithms are solved, I think every 'thing' in the universe h…

I think you would find metaphysics (specifically ontology) and cognitive science interesting.

I think you'll find your ideas are actually very, very old. ;)

Re: This AI Boom Will Also Bust

#264

Post by dquail »

dkarapetyan wrote: post #253
dquail wrote: post #245

Earlier quoted context omitted.

Agree generally. Except being unimpressed unless performance is achieved on sub Google scale hardware. Today's Google supermachine is tomorrow's raspberry pie. No need to artificially constrain our bounds. There is, after all, the inevitability of Moores law.

Moore's law has been dead for a while now. Most of the chip in your phone is powered off because otherwise it would burn up. Highly recommend watching this video: https://www.youtube.com/watch?v=_9mzmvhwMqw

Interesting. I've yet to hear a Moores law is dead argument, so perhaps I should watch the video before commenting further. But the fact that most of the chip is turned off, doesn't falsify the fact that most of it still exists. Cooling it properly is a separate problem independent of computation no?

Re: This AI Boom Will Also Bust

#265

Post by edblarney »

RandyRanderson wrote: post #102

I'm old enough to have seen a lot of these boom/hype/bust cycles. I'm convinced that this time is, in fact, different. To temper this I believe most decent user-visible changes will take ~5 years (as most actually useful software does) but the changes will be huge: * The author cites computer driven cars. I think this will take place mostly on long-haul highway trucking instead of in cities first. Even so, this could…

I think this is the best comment.

AI will find it's way into specific areas in which the ROI is very significant, and though those will be hard to predict, I agree with the above.

It will make a huge impact in some areas.

Others, not so much.

It's funny that not a single person mentioned 'big data'.

Big data was all the rage for the last few years, and it would seem like an intuitively obvious fit, not? I don't think so.

I think it's a stretch to see how AI help's the gap design new clothes, or even optimizes sales approach. I suggest it will be things like customer service as Randy indicated.

And yes, truckers will be the first to go.

Re: This AI Boom Will Also Bust

#266

Post by randcraw »

ffwd wrote: post #222
jonathankoren wrote: post #210

Earlier quoted context omitted.

This is literally a philosophical problem. It's called ontology. And no amount of data solves this problem, because ultimately it's a labeling problem, and the border between things is ill defined, and additional data doesn't help resolve labeling ambiguity, if anything it finds out just how ill defined the world actually is. Think about it. Let's say you had a problem which was find the black squares. So you collect…

> if anything it finds out just how ill defined the world actually is. Yeah I hear this but it seems only half-true to me. While for most intents and purposes the world is ill-defined, in another sense the world itself is "100% signal" and no noise. If we "zoom out" and take a grand view, imagining that we have a supercomputer and a huge database, and the algorithms are solved, I think every 'thing' in the universe h…

If I understand your point, I'd suggest that it may apply best to the use of DL for low level AI -- seeing, hearing/generating speech, and recognizing/ navigating/ interpreting complex signals of other kinds. There classification is secondary to modeling the many subsymbolic facets endemic to raw analog signals.

I suspect DL will eventually settle into a less vaunted role in the historical saga of AI than it portends now. And that role may well be the 'grounding' of sensory experience -- the modeling of the world into something perceptually and cognitively manageable, like Plato's shadows on a cave wall.

Re: This AI Boom Will Also Bust

#267

Post by jonathankoren »

randcraw wrote: post #266
ffwd wrote: post #222

Earlier quoted context omitted.

> if anything it finds out just how ill defined the world actually is. Yeah I hear this but it seems only half-true to me. While for most intents and purposes the world is ill-defined, in another sense the world itself is "100% signal" and no noise. If we "zoom out" and take a grand view, imagining that we have a supercomputer and a huge database, and the algorithms are solved, I think every 'thing' in the universe h…

If I understand your point, I'd suggest that it may apply best to the use of DL for low level AI -- seeing, hearing/generating speech, and recognizing/ navigating/ interpreting complex signals of other kinds. There classification is secondary to modeling the many subsymbolic facets endemic to raw analog signals. I suspect DL will eventually settle into a less vaunted role in the historical saga of AI than it portends…

This problem is deeper (HA!) than trying to apply a computer algorithm. It's a labeling problem. It's an interpretation problem. It's a human problem.

Re: This AI Boom Will Also Bust

#268

Post by danielrm26 »

I think a key differentiation between ML and more common statistics is that ML is designed to improve itself based on data. Statistical methods don't do that.

So maybe they're trying to do very similar things in a lot of cases, but self improvement is a major differentiator.

Re: This AI Boom Will Also Bust

#269

Post by randcraw »

pmrd wrote: post #227

Also - must remember that AI is a programmer on the sidelines letting data do the logic. It required a fairly different frame of mind than traditional programming

So data as employed by DL plays the same role that it does in Brooks' subsumption architecture -- grounding and shaping the knowledge model -- but now doing it emergently, albeit requiring a lot of parameter tuning from the human-in-the-loop.

An interesting prospect for the evolution of software developers.

Re: This AI Boom Will Also Bust

#270

Post by randcraw »

nwjtkjn wrote: post #38
toomuchtodo wrote: post #21

Earlier quoted context omitted.

> the people who'll be in trouble with be those without formal education to back them up. The people who can't hack it are those who'll be in trouble. Tech has never much been the place where credentials are necessary. Don't specialize and saddle yourself with years of college debt if you're unsure of the field's long term prospects.

FWIW, one does not accrue debt doing a PhD in something like Machine Learning.

The debt incurred in a wasted PhD is not monetary; it's lost time.
Post reply on HN