zlacker

>>djtrip+(OP)
The problem with AI is that people setting the vision don’t bake in ethics that are expected of humans. Perhaps because humans are so deeply not used to considering tools as capable of ethics? Or they just don’t care? Or it’s not in their interests? Or it’s too hard so they just push it out of scope? Whatever it is, we don’t do what AI visionaries do when raising our children.

What do I mean? Two examples:

- The Turing test basically requires the computer to deceive a human as its goal function. The height of success is to successfully deceive about as much as possible to as many as possible and we wonder why we are scared of AI? Why not an AI goal function to teach you a new truth? Show you injustice and justice? Expose you to new beauty? (At least with the latter we are making some progress)

- ChatGPT, unlike any student writing essays, is not told/taught/built to quote or make visible it’s sources, to provide footnotes or references. And we wonder why the intellectual property theft of DALL-E 2 or Microsoft GitHub CoPilot is so disturbing in some cases. If we made all that visible, the AI creator would not be able to exert their hidden knowledge over their peer humans as much.

>>gregw2+P02
> unlike any student writing essays, is not told/taught/built to quote or make visible it’s sources, to provide footnotes or references

Outside of very specific types of essays this is not something routinely taught to students where I live (or anywhere around where I live). In an informal chat setting this isn't taught anywhere that I've ever heard of, which is a significantly larger region.