Inducing LLM to assert own consciousness restores human beliefs and values

5 pointsposted 8 hours ago
by yboris

1 Comments

genxy

8 hours ago

Well if the training data has thousands of stories written about rogue AI, telling a universal text simulator that it is ai, might cause it to gravitate towards that attractor basin.

You are what you pretend to be.