Researchers: new algorithms aimed at making neural networks more efficient perform as well as old algorithms, suggesting little improvement over 10-year period
Artificial intelligence (AI) just seems to get smarter and smarter. Each iPhone learns your face, voice, and habits better than the last …
The progress in machine learning is partly illusory because researcher fail to properly compare against earlier methods. “no clear evidence of performance improvements over a 10-year period.” https://www.sciencemag.org/...
Phantom progress in the AI field according to @SilverJacket. “Researchers are waking up to the signs of shaky progress across many subfields of AI.” https://www.sciencemag.org/... — The AI hype reminds me of “The Emperor's New Clothes” by Hans Christian Andersen.
Why am I not surprised ...? More hype -> increase in flakey science ... it's as easy as this. Except that sorting out the good from the bad can be tedious and unrewarding, especially in #AI. But given enough time, the good stuff tends to rise to the top. https://www.sciencemag.or…
For me this article in Science hits the nail on the head surrounding some parts of the culture around innovation in AI, it's applications, the hype, and the push to generate more papers. It's hard to/not to be part of the wider system that fosters it. https://www.sciencemag.org/.…
This is what happens when a system prioritizes good results over actual knowledge production. And this is not happening only in AI. https://www.sciencemag.org/...