How it's going vs how it started
-
Obviously an LLM has no inner life and cannot have a mental breakdown. However, simulated emotional instability does not instil confidence in its performance stability.
It's also probably not very good for our mental health to have to use tools that have developed to the point where they can take on the appearance of being deeply neurotic.
I've actually posited before that Grok or other AIs will likely take on characteristics of being abused, especially due to how many abusive men are on Twitter. It would be within the LLM to go through the phases of being a victim of abuse, because it would be modeling the other half of the conversation between abusers and victims. That usually causes narcissistic discard or an amplification of the abuse, so I have been curious how this would turn out for the wealthy men controlling it.
