Musings on AI

I’ve done a ton of reading about artificial intelligence over the years: Marvin Minsky, Ray Kurzweil as well as nanotech pioneer David Drexler.  Machine consciousness, by my lights, is still at least half a century away. What people are hooping and hollering about and calling ‘AI’ is actually expert systems and machine learning. A dear friend whose IQ is considerably above my own had this to say:

“Yeah, I don’t know why the people don’t know it’s pointless crap!! I suspect a lot of the compsci people are confused about the differences between natural human language and the artificial programing languages we create. Real languages are very complex and always changing, and every speaker/hearer has a similar but different model. Language without context usually doesn’t mean much. They should be saying psychotic break* rather than hallucinations, though both would be gross mislabeling of what is actually happening, there’s not enough Intelligence or understanding there to be psychotic!

Those connected to it are clearly having a future earnings high unconnected to reality. I have a lot less respect for nvidia than I did before this crap! We’re both old enough and smart enough to know that most innovations are more hype than progress. Those working on LLM need to work on their’ internal LLM. No one learns language without out a lot of interaction with those who already know how the language works, mostly, and real world outcomes shaped by their’ own internal model. It’s called statistics, and they are one of the best ways to confuse reality in human minds, or machines. I guess they think of themselves as pretty “Hep cats”… and intelligent… it’s much worse than the computers in 1984 that wrote fiction for the ignorant masses. Too many smart people are way to supportive of this notion.

It’s a fine example of what’s wrong with the stock market and traders that people are making huge, huge monetary commitments to this crap. Not to mention these same morons saying AI’s need for power is a good reason to use nuclear power. LLM’s will not save humanity, lead to scientific break throughs, or lead to better understanding by anyone other than those who’ve wasted time and effort on it. Way too many people acting on incomplete information they don’t understand.”

Which sums it up very nicely.

Step back folks, read a little and educate yourself before you start trusting this new trend.

We were discussing this Nature article: AI models fed AI-generated data quickly spew nonsense