If “AI” was really half of what it was cracked up to be, here’s how I would imagine it would have gone down:
AI Co: “Hello everyone. We would like to present our AI code writer. “AI, please write the code behind Google’s search engine.”
AI: “Here ya go: <splort>”
Google: “That’s very impressive.”
AI Co: “AI, please fix all the bugs in Google Voice that make it run slow as dog feces.”
AI: “Done. <splort>”.
Google: “Ok, we will give you $50B right now for 100% of everything you own. Don’t breathe a word of this to anyone.”
6 months later: “In today’s news, Google Inc. has just released it’s 30th major product. The industry is shocked and suspects perhaps Google has figured out some type of AI program to write code. Google has announced it has purchased AI Inc. for an undisclosed amount. Google has also announced it is laying off 60% of its staff worldwide. Only CEOs and marketing personnel remain.”
In reality, Sam Altman appears to be a man who has created a machine that allegedly points out massive gold deposits. But rather than mining gold, he insists on selling machines that find gold mines.
I recently also said basically the same thing. If it was actually AI you could just describe an app and it would spit out a polished finished program in minutes. Instead it can write a few blocks of code at a time maybe, that might be complete garbage code or might work fine, total toss up.
Many state that the “AI hallucinates 30% of the time”. I really appreciate the observation of someone else online who observed, “No, it hallucinates 100% of the time, only sometimes the output happens to be useful.”
Yeah that’s a good point. For it to be useful at all, I think that frame of reference is necessary