Ooh ooh ooh!! I want a Skynet too!!!
---
Speaking of robot AI, I figured the most effective way to get to closer something actually non-laughable is to teach them communally, where several independent units can communicate with each other. You would need a scenario where they must test and try different things, and different things provide either 'pleasure points' or 'pain points'. (Mmm, pleasure points...) Then, when one learns something, it can communicate these things, but that doesn't mean others will believe it. They must test it for themselves and if the 'teacher' was right, it earns 'trust points' with others. Some robots would gain pleasure points from saving others from pain, some from directing towards pleasure, but some would need to be intentionally programmed backwards in this regard. This would require each individual to learn how to trust and validate information, and whoever becomes the best at this and be a winner. I wanna be a winner!
...Umm, or something...
No comments:
Post a Comment