This week, an anonymous chap ( bear with me ) posted their theory of cognition on one of my groups. We had a brief exchange, but the main feedback I had for them was this :
"You cannot build a correct model of cognition, without first locking down your underlying sensory data types. You simply can't get started. Sensory data is the underlying data upon which operations are defined, and there are only two things in computation : data, and operations on data."
I post this diatribe, yet again, because it captures so much of how AI has missed the boat repeatedly in the 60s, in the 80s, and so-far ... in the 2020s. The eventual solution will be remarkably straightforward, because the missing parts are so straightforwardly missing.
We just have to wait for systems to be built utilising such basic typing.
No comments :
Post a Comment