41 ms·
At the risk of sounding like a broken record I must point out that AI, neural networks, machine learning, Perceptrons, algorithms and deep learning are not inte
by YeGoblynQueenne 7y ago
At the risk of sounding like a broken record I must point out that AI, neural
networks, machine learning, Perceptrons, algorithms and deep learning are not
interchangeable terms, any more than "maths" is interchangeable with
"calculus", "trigonometry", "geometry", "algebra" and "arithmetic".
Unfortunately this article is typically using such terms as synonyms and,
while it's probably meant to help the lay reader understand a technical
subject, will only manage to confuse the issue further.
It's clear that there is intense interest in deep learning and machine
learning in the last 7 years or so, but, especially when talking of "AI", it
is important to remember that the majority of work in AI has _not_ been on
machine learning, but on inference and reasoning, and using symbolic,
logic-based techniques (including Bayesian reasoning, as in Solomonoff's
Inductive Inference) rather than "statistical" methods (in the sense that the
term is used in AI to mean everything from probabilities to calculus).
Finally, there are historcial inaccuracies in this article. For example,
Rosenblatt's Perceptron was not "the first machine-learning algorithm' as the
article says: Arthur Samuel's self-playing checkers agents and Minsky's SNARC
are two systems predating the Perceptron by a few years and research on
machine learning techniques like Markov chains goes back to the 1910's.
- chobeat 7y agoWhile agree with you on the need to straighten the narrative and usage of these terms, the author is a philosopher and uses AI not in the common sense nor in the Computer Science sense. It is indeed confusing but he's very very rigorous in the way he writes (this is extremely simple writing for his usual style) so don't believe he's using the terms intercheangeably
- whatshisface 7y agoMost of these philosophical thinkpiece authors are charlatans. Charlatan philosophers can get along very well, even in professional circles, because there's almost no way to tell the difference between a true belief and a false belief in some areas of philosophy. However, they meet their downfall in articles like these, when they make the mistake of talking about something that is relatively easy to verify - revealing that they are sloppy thinkers with great English skills.
- chobeat 7y agoI don't get what you're talking about. I'm a Machine Learning engineers and I find his work one of the best being conducted on the topic. You can say he's snob or inaccessible, but for me Pasquinelli is in the top 1% of the tech critique, especially in the theoretical part. Also he's not even a native speaker and I don't find his way writing very compelling. Quite the contrary. Why do you think he's a charlatan?
- YeGoblynQueenne 7y agoI don't know anything about the author of the piece. I understand that it's a sketch of a larger article that he intends to write for publication. I hope that he will have the time to refine his scholarship until then. I'm not sure what to do of the ideas discussed in the article, themselves. I am honestly confused about what he means when he's talking about "AI", given his loose use of terminology, so I'm not sure I understand the article clearly enough to agree or disagree with it. To clarify, I'd be all for reading a strong critique of the modern (last decade) use of machine learning, including neural networks and deep learning (but not only). If he's talking about AI as a whole, however, I don't see how his critique can really work. I don't think I've ever heard of anyone using a classical planner or a constraint solver to drive an automated truck, say. The start of the article, with the discussion of the Agnicayana ritual was very intersting.
- chobeat 7y agoTo clarify how he uses AI, I suggest you to read this: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3078224 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3078224 This might help you to bridge your understanding of AI with how it's used elsewhere.
- empath75 7y agoYeah, of the things that he writes about that I know of, he seems to not quite get them right, but this is still an interesting article with a lot of big ideas to think about.
- whatshisface 7y agoThe curious case of the author that sounds stupid when you know what he's talking about and visionary when you don't. ;)
- empath75 7y agoI don’t think he sounds stupid at all — I think he’s just bending things to fit. People can be wrong and still be interesting.
- chobeat 7y agoIf he sounds like he's bending things, that should sound good to your ears. Otherwise he would be just reinforcing what you already believe and how you already see the world and therefore not doing any good philosophy.
- DATACOMMANDER 7y agoI know you’re being sarcastic, but I’ve integrated this idea into my bullshit detector. When I’m dealing with someone who purports to be an expert in a field I know little about, I will try to steer the conversation towards things that I know a lot about—without disclosing that I know a lot about them. People will show you who they are if you let them.
- chobeat 7y agoYou should read some of his more rigorous works: http://matteopasquinelli.com/ http://matteopasquinelli.com/