3 ms·

The AGI hype train is running out of steam

mindcrime 5y ago
LOL... what "AGI hype train?" There's been very little mainstream interest in AGI for the last decade plus. The hype train has been all around Deep Learning, mostly for image recognition, speech recognition, facial recognition, etc. The people working on AGI are quietly toiling away in their little corners of the world, working with very little regard for what Larry Page, Elon Musk, Peter Thiel, et al think. And most of them don't even self-identify as working on "AGI" exactly because the term is so often associated with fringe-science and outright quackery. Most of people doing AGI probably identify as working on "cognitive architectures", "cognitive psychology", or "computational linguistics" or some other field which is closely related but not "AGI" per-se. Hyping AGI helped spurred enormous investment in artificial intelligence, but overconfident predictions could be detrimental to progress in the field — if the real-world advances prove disappointing. WTF? Nobody has been hyping AGI. Most of the people talking about AGI lately, to the extent that anybody has been talking about it, have not been people actively working in the field. For comparison, look at the Google Trends view of the search terms "artificial general intelligence" and "deep learning" over the years. Note: I used "artificial general intelligence" spelled out and not the abbreviation "AGI" because "AGI" has many alternate meanings including "Adjusted Gross Income" which seem very popular and add noise to the results. https://trends.google.com/trends/explore?date=today%205-y&q=%22artificial%20general%20intelligence%22,%22deep%20learning%22 https://trends.google.com/trends/explore?date=today%205-y&q=... Scientists are also increasingly skeptical. A recent study paper posited that AGI is “in principle impossible,” while other researchers have mocked the term’s proponents. One scientist, who appears to not be an expert in the field, publishing a paper in Humanities and Social Sciences Communications, which merely resurrects outdated arguments from Herbert Dreyfus, and which conflates "human level general intelligence" with "human like intelligence and behavior in all areas" is hardly convincing, IMO. In the book he argued that an important part of human knowledge is tacit. Therefore, it cannot be articulated and implemented in a computer program. See, this is bollocks. All this argues is that you can't take that tacit knowledge, articulate (eg, encode) it and hardwire it into a computer program. But it is not a proof, or even a statement to the effect that, you can't give the program adequate means to interact with the world and then let it learn this same knowledge on it's own. There's also an unsupported assumption here than an AGI needs all of the tacit knowledge that a human has. All of that said, the AGI Conference 2021 just happened, and anybody really interested in what's happening in the AGI world is welcome to watch all of the sessions for themselves. https://www.youtube.com/watch?v=FBwiURdT5rI&list=PLvcs9Kg3xCIrjkXvpQpJbdClFOM6mQ9f5 https://www.youtube.com/watch?v=FBwiURdT5rI&list=PLvcs9Kg3xC...
- kordlessagain 5y ago
  Agree!

LOL... what "AGI hype train?" There's been very little mainstream interest in AGI for the last decade plus. The hype train has been all around Deep Learning, mostly for image recognition, speech recognition, facial recognition, etc. The people working on AGI are quietly toiling away in their little corners of the world, working with very little regard for what Larry Page, Elon Musk, Peter Thiel, et al think. And most of them don't even self-identify as working on "AGI" exactly because the term is so often associated with fringe-science and outright quackery. Most of people doing AGI probably identify as working on "cognitive architectures", "cognitive psychology", or "computational linguistics" or some other field which is closely related but not "AGI" per-se. Hyping AGI helped spurred enormous investment in artificial intelligence, but overconfident predictions could be detrimental to progress in the field — if the real-world advances prove disappointing. WTF? Nobody has been hyping AGI. Most of the people talking about AGI lately, to the extent that anybody has been talking about it, have not been people actively working in the field. For comparison, look at the Google Trends view of the search terms "artificial general intelligence" and "deep learning" over the years. Note: I used "artificial general intelligence" spelled out and not the abbreviation "AGI" because "AGI" has many alternate meanings including "Adjusted Gross Income" which seem very popular and add noise to the results. https://trends.google.com/trends/explore?date=today%205-y&q=%22artificial%20general%20intelligence%22,%22deep%20learning%22 https://trends.google.com/trends/explore?date=today%205-y&q=... Scientists are also increasingly skeptical. A recent study paper posited that AGI is “in principle impossible,” while other researchers have mocked the term’s proponents. One scientist, who appears to not be an expert in the field, publishing a paper in Humanities and Social Sciences Communications, which merely resurrects outdated arguments from Herbert Dreyfus, and which conflates "human level general intelligence" with "human like intelligence and behavior in all areas" is hardly convincing, IMO. In the book he argued that an important part of human knowledge is tacit. Therefore, it cannot be articulated and implemented in a computer program. See, this is bollocks. All this argues is that you can't take that tacit knowledge, articulate (eg, encode) it and hardwire it into a computer program. But it is not a proof, or even a statement to the effect that, you can't give the program adequate means to interact with the world and then let it learn this same knowledge on it's own. There's also an unsupported assumption here than an AGI needs all of the tacit knowledge that a human has. All of that said, the AGI Conference 2021 just happened, and anybody really interested in what's happening in the AGI world is welcome to watch all of the sessions for themselves. https://www.youtube.com/watch?v=FBwiURdT5rI&list=PLvcs9Kg3xCIrjkXvpQpJbdClFOM6mQ9f5 https://www.youtube.com/watch?v=FBwiURdT5rI&list=PLvcs9Kg3xC...

kordlessagain 5y ago
Agree!