Richard Stallman’s positions on software seem less extreme every year.
Ask 1990s or 2000s me, and I’d say he was interesting but “out there”.
Now, I’m thinking he was just 30-40 years ahead of his time.
I see the big fight now is over AI. The talk is over AI ethics and safety … but can we honestly entrust that to Big Tech?
The reality is they simply cannot, and won’t. While safety will (probably) be assured, because unsafe products bring lawsuits, unethical products that extort customers for money are apparently perfectly legal. We see this in pharmaceuticals, processed foods, and of course Big Tech.
But requiring licenses to construct or deploy AIs would be a sufficiently large moat.
Is there a lesser of two evils here? I honestly don't know, but I don't think we will get a choice.
I should have control over model updates, retraining, the training pipeline, the prompts... Everything. How ethical or safe the ai is is my problem. It's not a matter of trust, but that it gets changed out from under me, breaking my workflows, and not letting me fix problems with it that affect me.