It happened to VR a few years ago. The development slowed down when companies realized the sales. Unlike VR, AI is not a physical good that users/customers can purchase and test directly. It is just a buzzword. Companies will use it to death at every opportunity to market themselves. This is unstoppable. We will hear more about it, maybe until people get sick of hearing "another crappy ai product".
An Epidemic of AI Misinformation
21–30 of 68 posts
Re: An Epidemic of AI Misinformation
#22The buzzword back in the 70s-80s (after AI over-promised in the 1960s) was 'expert systems'. https://en.wikipedia.org/wiki/Expert_system
(To the extent that I have kept up with it) modern AI skips the 'knowledge base' part of ES, in favor of pattern-recognition based on 'training'.
Today's (Indeterministic, trained, n-net) AI has clearly saved a lot of time/effort in creating 'knowledge bases'. I suspect it appeals more to singular fantasies about 'more human than human' intelligence. (Sorry Ray)
Question is: Is today's AI even a magnitude-better than (deterministic) ES insofar as extensibility and verifiability? What if we had spent those decades refining the ES approach instead?
Re: An Epidemic of AI Misinformation
#23OpenAI hype is definitely exaggerated. Seems like there might be a connection with Elon Musk somehow because all of his projects seem to get a massive amount of hype also.
The radiology thing, is it really the case that there are no startups in that area with useful AI software? Seems like he overstated that.
Part of this is a worldview difference. Many people truly believe that AGI is just around the corner, and that even before then, the narrow AI applications will significantly alter the world as they are deployed. And since no one can actually predict the future, it's an area where it's easy to have different worldviews.
Personally I think that it's true that there is a lot of poor reporting and companies that overhype results, but it also seems like people like Gary Marcus are really not keeping up to date with the true capabilities of DL systems. If he was up to date, why would he be so pessimistic about applications like radiology? There seem to already be a lot of strong results.
Re: An Epidemic of AI Misinformation
#24Very hard to critique an article you overwhelmingly agree with! The key point I kept picking up was the extent to which a press willing to laud a discovery was reticent about owning the clinb-down. Peer review in ML journals should be tighter maybe? If you solve a limited subset of the three body problem you can't claim to solve "the three body problem" and if you apply a well known Rubik's cube model solution you di…
typically the actual papers are reasonably honest about everything and the misinformation comes in elsewhere, often university press releases summarizing the results.
Re: An Epidemic of AI Misinformation
#25The buzzword back in the 70s-80s (after AI over-promised in the 1960s) was 'expert systems'. https://en.wikipedia.org/wiki/Expert_system (To the extent that I have kept up with it) modern AI skips the 'knowledge base' part of ES, in favor of pattern-recognition based on 'training'. Today's (Indeterministic, trained, n-net) AI has clearly saved a lot of time/effort in creating 'knowledge bases'. I suspect it appeals m…
They're not even competing with each other. How would you compare an ES to a deep neural network? According to what benchmark? What do you mean by verifiable? Modern success stories with machine learning often deal with problems (NLP, vision, RL) that have little to do with the problems that were being solved by expert systems.
Also, an hypothesis: if (big if) you somehow managed to have an expert system outperform deep learning for vision, I bet that it won't be any more verifiable than a deep neural net is today.
To me the complaint that modern deep learning is unverifiable is a bit dumb, in the sense that any perception algorithm working with low level signals (vision, sound) will not be transparent to a human. 15 years ago, an image classification pipeline looked like: bag of SIFT features + SVM classifier. Try explaining the decision made by that algorithm in an intuitive way!
Re: An Epidemic of AI Misinformation
#26It happened to VR a few years ago. The development slowed down when companies realized the sales. Unlike VR, AI is not a physical good that users/customers can purchase and test directly. It is just a buzzword. Companies will use it to death at every opportunity to market themselves. This is unstoppable. We will hear more about it, maybe until people get sick of hearing "another crappy ai product".
AI is more of a datacenter technology. People already use it (siri,alexa,google search) and they apparently like it already. It was easy to see that any VR headset is bulky and agonizing, but those new services work fine. Sure, however the "not hot dog" apps are going to be disappointing
Re: An Epidemic of AI Misinformation
#27Let us remember that AI is at the center of the cloud wars. Google, Amazon and Microsoft while they produce ton of research, their sales teams need to market those R&D investment to get customers interested in their tech. We are at a stage that AI is still a buzz word for many companies, once we evolve and deploy more and more of AI cases, we will see less and less articles promising things that can't be implemented
That's probably a part of it. Note how the Google TPUs aren't for sale. If you want them you have to use the Google Cloud. The cloud is expensive and slow ... I think everyone is shocked when they first see perf numbers coming off Azure. I don't know if GCE is better, but the temptation to overload the hardware is always there: hardware rental is fundamentally a business with low barriers to entry. Anyone can buy som…
I work for a company that uses bare metal, open stack, azure, AWS, and GCE.
GCE perf is significantly better (and more consistent) than Azure. Even with Windows instances. :/
But I agree that “cloud is slow” when compared to bare metal- it’s also most financially costly especially for the same performance due to it being slower. But the gains in flexibility are immeasurable.
Re: An Epidemic of AI Misinformation
#28Ugh. I dislike the author s permanent negativity, but he s right about a lot. I think it’s worth asking why people feel the need to lie about the future of AI? If they are confident about its future (and I don’t know of a fundamental reason why they would not be) then there is no reason to rush half assed results out the door and overcompensate (like gpt2). There is plenty of theoretical questions and answers to deba…
It's the bullshit process by which people get funding. Make something that has some potentially interesting engineering value, hype it up to hell and back, and funding agencies, investors etc. are more likely to back you. Probably this wasn't everyone's first plan all along, but when people who make a marginally better hot-dog / not-hot-dog classifier put so much spin on it, then everyone else has to just to remain v…
> impacts so much of academia
i think the trend started from the industry , but you re right warrantless self-promotion very pervasive in academia , and it's sad that it works!
Re: An Epidemic of AI Misinformation
#29The buzzword back in the 70s-80s (after AI over-promised in the 1960s) was 'expert systems'. https://en.wikipedia.org/wiki/Expert_system (To the extent that I have kept up with it) modern AI skips the 'knowledge base' part of ES, in favor of pattern-recognition based on 'training'. Today's (Indeterministic, trained, n-net) AI has clearly saved a lot of time/effort in creating 'knowledge bases'. I suspect it appeals m…
I mean, you extend a DL system by giving it examples. Yes, in fact by harvesting example or annotation data you can extend models many magnitudes faster and more comprehensively than through manual analysis. Effectively the analysis is done automatically. At this point, what comes out it is usually poorly factored and entangled and not human-parseable, but that doesn't prevent it from being applied effectively to a range of narrow tasks that expert systems cannot approach.
We do still rely on expert systems for things that we want to be carefully parsed, verified and analyzed by people though. Such as rules that oversee most self-driving cars based on perception handled by neural networks. However, not all self-driving systems lean as heavily on rules at the top level as others.
Re: An Epidemic of AI Misinformation
#30> I see this as a version of the tragedy of the commons, in which (for example) many people overfish a particular set of waters [...] if and when the public, governments, and investment community recognize that they have been sold an unrealistic picture of AI’s strengths and weaknesses that doesn't match reality, a new AI winter may commence.
I think this is not only inevitable, but necessary! This time around it has been a lot more useful, due in no small part to the advance in hardware since 1970s.
Unfortunately this has caused many important people to believe far too much of the hype and not see it's current limitations. As a result they have started integrating it into important part of our societies - i find this alarming - not for the reason most people find it alarming i.e "because it's too smart", but because it's far far too dumb in combination with people assuming it's very smart. I think a lot of this problem stems from inappropriately anthropomorphising ML with terms like "AI" when we are no where near the stage that we need to have the philosophical debate about where something is sentient or "intelligent". The ML we are doing with NNs is still at the "tiny-chunk-of-very-specifically-engineered-piece-of-brain" stage. It's important people understand this before we start integrating what are essentially basic statistical mechanisms into our societies.
For those in pursuit of better ML and things like real AI aka AGI, I also think having the hype blow away will do more good in the form of clarity and lack of noise than it will harm in lack of funding.