Alt Text: A comic in four panels:
Panel 1. On a sunny day with a blue sky, the gothic sorceress walks away from the school with the Avian Intelligence Parrot in her hands toward the garbage.
Gothic Sorceress: “Enough is enough, this time it’s straight to the garbage!”
Panel 2. Not far away, a cute young elf sorceress is discussing with her Avian Intelligence in the foreground. Her Avian Intelligence traces a wavy symbol with a pencil on a board, teaching a lesson.
Elf Sorceress: “Avian Intelligence, make me a beginner’s exercise on the ancient magic runic alphabet.”
AI Parrot of Elf Sorceress: “Ok. Let’s start with this one, pronounce it ‘MA’, the water.”
Gothic Sorceress: ?!!
Panel 3. The Gothic Sorceress comes closer and asks the Elf Sorceress.
Gothic Sorceress: “Wait, are you really using your?!”
Elf Sorceress: “Yes, the trick is not to rely on it for direct answers, but to help me create lessons that expand my own intelligence.”
Panel 4. Meanwhile, the AI Parrot of the Elf Sorceress continued to write on the board. It traced a symbol of poop on the board, then an XD emoji. The Gothic Sorceress laugh at it, while the Elf Sorceress is realizing something is wrong with this ancient magic runic alphabet.
AI Parrot of Elf Sorceress: “This one, pronounce it BS, the disbelief. This one LOL, the laughter.”
Gothic Sorceress: “Well, good luck expanding anything with that…”
GenAI doesn’t “know” anything. A 15 year old who spends a year copying his friend’s physics homework will learn a tiny bit of physics. GenAI is just generating something new without actually learning information.
It’s a fancy auto-complete that looks at the entirety of human writing and guesses what word should come next based on statistical probability. That isn’t learning, that’s rolling dice 10,000 times and seeing what number comes up most often.
GenAI cannot “intend” anything. It cannot develop consciousness any more than Akinator or a Tickle-Me Elmo can. The correct way to handle this technology is to treat it with reality: as a tool that can quickly look at a lot of stuff and not as a developing mind.
I’m well-aware of all that, but if you think that’s not going to change, you’re a bigger fool than the AI-evangelists. Even if it doesn’t change, the distinction won’t matter all-too-soon.
By all means, leave the overblown toy to the delusional right-up-until AI, whether truly intelligent or better-at-faking-it than today, has killed us all.
I think of it as outsourced intuition. It provides a first gut feeling response to the question based on what the Internet would say. That can be useful if you need a starting point. It very rarely should be an ending point.