Thats the current pivotal question in the AI community. A video well worth your time.
I personally think that the conversation will turn to the quality of data having a more profound effect on AI performance. What goes in dictates what comes out.
Lets take history as the ultimate example. We know that history is written, then rewritten as required by the powers that be (TPTB).
“Who controls the past controls the future. Who controls the present controls the past.”
George Orwell
I dont think its hard to see that AI will be intentionally (mentally) handycapped by what “data” TPTB allow AI to be trained upon. They already have “guardrails” to limit where AI may go. And then there are Usage Policies that just happen to include limits on Law, Medical and Financial, but most importantly, Hate. How convenient. Now we wouldnt want to allow just any old “data” that might offend someone, would we?
That means that historic “data” is not accurate, and therefore AI preformance will not be accurate. In fact AI may turn psychotic because the “data” is schizophrenic. You know where Im going with all of this.
I think we’ll stick with the current human trajectory of willful ignorance instead of enlightenment.