Yes, AI (or even better what I like to call Artificial Stupidity - naive implementations that get you 90% there) are a great cognitive prosthetic, but in my experience should not be relied on.
What is intelligent about natural pattern recognition? /s
Sarcasm aside, your definition is too low-level, as if we called brains just braincells. Pattern recognition is a building block of AI like a logic gate is a building block of a CPU.
If you studied formal logic, you understand what logic gates do and what a CPU can do.
We don't understand brains or braincells - how many operations per cecond can a braincell do? What type of operations? How many bytes does it store? We don't know.
Can we replace a single braincell with a chip and get them to act the same?
Likely not, but that was not my point either, it was just an non-stretchable analogy.
AI is not a single network usually, it is a pipeline of networks of different kinds, see e.g. https://m.youtube.com/watch?v=P1IcaBn3ej0 (seek to 1:35 for beginning of an architecture overview). You can’t call it “just pattern matching” without fetching it too far.
I have been using the phrase "Artificial Stupidity" as well, but with the opposite meaning. Specifically I like to think of human-like artificial stupidity as a challenge for machine intelligence, in which an algorithm is able to replicate the rather sophisticated and incredibly entangled logic, intuitions and calculus of humans at the height of their stupidity. This seems to me a much greater challenge than the standard sort of supervised learning problems in that a truly stupid AI must be able to imagine latent variables that allow it to explain away real world observations in a way that is both statistically implausible but casually serendipitous to their stupid peers. This seems to me to be a requirement for any kind of useful AGI.
Github Copilot is a great example of this IMO. When you're writing code it seems to be on a 90/10 of 80/20 split where 10-20% of your code is the actual important bits that you have to pay attention to and the other 80-90% is all the common patterns, components and conventions around the important code.
AI is fantastic for the 80-90% of that code that I really don't care about. I can focus my energy on the 10-20% of the code that is really important and let the AI code the other 80-90% that is mostly glue and filler.
reply