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

Stop Calling Everything AI, Machine-Learning Pioneer Says

spectrum.ieee.org

Post reply on HN
Print view View original

71–80 of 123 posts

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#71

Post by sgt101 »

72deluxe wrote: post #24
drpixie wrote: post #4

Michael I. Jordan is spot on. We have NO artificial intelligent systems, for any sensible definition of "intelligent". None. ZERO. We have "big data". We have (brittle and rather poor) "pattern recognition". We have (very limited) "simulate a conversation". But we do not even know what "intelligence" really means! We (the industry) should recognise "AI" as term that has become essentially meaningless. "AI" is now not…

Correct. Apparently my phone has "AI" because it recognises a flower as a flower and applies a colour filter and background blur when I use my camera. This is not AI. By the same extension of logic, any program that recognises input data and performs some form of pre-programmed logic is AI. ie. any computer program?

Recognizing a flower is absolutely AI.

https://xkcd.com/1425/

ok - or a bird.

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#72

Post by dr_dshiv »

The term Artificial Intelligence implicitly makes people design systems that don’t involve people—because if people have an information processing role, then it seems like the system isn’t artificial yet, and therefore unfinished. That’s rather unhealthy

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#73

Post by blacksmith_tb »

joconde wrote: post #62
jvanderbot wrote: post #20

I'm in the submarine AI camp[1]. I'd rather build tools that kinda help humans do thinking, rather than thinking machines. Like how submarines help humans swim but who cares if they are actually swimming. 1. "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra

> 1. "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra Not sure what he meant by that, it doesn't really make sense to me: thinking yields information and deductions that can be very useful, while swimming by itself accomplishes nothing useful. Actually, I just can't make any sense of the sentence, because while I have a general…

Not a polymath physicist, but I take Djikstra to be saying "don't obsess so much over the differences you fail to see the similarities". The submarine doesn't swim, but it still gets somewhere in the water.

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#74

Post by bitwize »

Today's AI is, as I like to call it, statistics with sexy marketing. A lot of AI programming consists of loading Python modules that correspond to various models, and fooling around with them to see which best fits the data you have. In other words, you're working more like a mathematician experimenting with potential solutions.

There's pressure at my job for architects to "leverage AI". What I always suggest to them is to find a statistician and see if things like neural networks are even necessary before committing to them. Sometimes the problem could best be solved with a heuristic, a rules engine, or a simple statistical model like linear regression, in which case "leveraging AI" is merely hype.

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#75

Post by lkschubert8 »

joconde wrote: post #62
jvanderbot wrote: post #20

I'm in the submarine AI camp[1]. I'd rather build tools that kinda help humans do thinking, rather than thinking machines. Like how submarines help humans swim but who cares if they are actually swimming. 1. "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra

> 1. "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra Not sure what he meant by that, it doesn't really make sense to me: thinking yields information and deductions that can be very useful, while swimming by itself accomplishes nothing useful. Actually, I just can't make any sense of the sentence, because while I have a general…

The point is the goal of a submarine is to allow humans to traverse large distances underwater, the semantics of how it does so are unimportant. Similarly the ability of a computer to think is moot, its the results we get from that that matter.

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#76

Post by DonHopkins »

sgt101 wrote: post #71
72deluxe wrote: post #24

Earlier quoted context omitted.

Correct. Apparently my phone has "AI" because it recognises a flower as a flower and applies a colour filter and background blur when I use my camera. This is not AI. By the same extension of logic, any program that recognises input data and performs some form of pre-programmed logic is AI. ie. any computer program?

Recognizing a flower is absolutely AI. https://xkcd.com/1425/ ok - or a bird.

Not hotdog!

https://www.youtube.com/watch?v=vIci3C4JkL0

https://apps.apple.com/us/app/not-hotdog/id1212457521

https://play.google.com/store/apps/details?id=com.codylab.se...

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#78

Post by mirker »

blacksmith_tb wrote: post #73
joconde wrote: post #62

Earlier quoted context omitted.

> 1. "The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra Not sure what he meant by that, it doesn't really make sense to me: thinking yields information and deductions that can be very useful, while swimming by itself accomplishes nothing useful. Actually, I just can't make any sense of the sentence, because while I have a general…

Not a polymath physicist, but I take Djikstra to be saying "don't obsess so much over the differences you fail to see the similarities". The submarine doesn't swim, but it still gets somewhere in the water.

[And a program/computer may not think yet still does interesting computations]

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#79

Post by mathematically »

IshKebab wrote: post #52
mathematically wrote: post #2

Unrelated to the article but I ran across an interesting result recently that is related to AI (and the hype surrounding it): Let A be an agent and let S be a Turing machine deductive system. If A correctly understands S, then A can correctly deduce that S does not have A’s capability for solving halting problems. [1] One way to interpret this is that all existing AI systems are obviously halting computations simply…

> all existing AI systems are obviously halting computations simply because they are acyclic dataflow graphs No they aren't. Think about LSTMs for example.

So how do you get a value out of an LSTM?

Re: Stop Calling Everything AI, Machine-Learning Pioneer Says

#80

Post by 6gvONxR4sf7o »

rspoerri wrote: post #8

Interestingly as soon as a system works or we understand it's principles we stop using the term "artificial intelligence". It becomes image recognition, autonomous cars, or face detection. We mostly talk about artificial intelligence when we dont really know or understand what we are talking about :-)

There's more nuance here than is usually given credit. I think it's more that once we understand the principles and the system works, we realize it never really needed AI. Here's how the story goes:

Someone comes along and poses a problem. It seems like an automated solution will require building something like R2-D2. In other words solving it will need AI. Then someone else comes along and finds a way to solve it that looks nothing like R2-D2. Maybe they solve it with a massive dataset and a GLM. Turns out it never really needed AI.

As a field, we're prone to thinking R2-D2 is necessary and sufficient for a task, when it keeps turning out it's a sufficient condition but not a necessary one for so many tasks.

Post reply on HN