Half of my custom system prompt for ChatGPT is me trying to solve this problem by telling it to challenge me, that I value the truth, and that I enjoy being told that I'm wrong.
Some things still slip through, but overall it's helped attenuate those kinds of behaviours
so, we’re now entering the polytheistic era - the time of many flawed and capricious gods, like the Greek or Norse pantheons. fair to assume that AGI will bring the one true God that sweeps them all away?
If you're interested in the dynamics of people perceiving AI to be conscious, check out Robert Saltzman's Substack ( https://robertsaltzman.substack.com/ ) and his book Understanding Claude ( https://www.amazon.com/dp/B0G5L9F998/ ). Saltzman has probed deeply into the "mind" of Claude and the minds of Claude users and offers compelling explanations.
Re: “Your empathy circuits are being hacked”
I recently wrote to a friend that AI had certainly mastered retention: all it had to do was constantly say how insightful my prompts and queries are.
Half of my custom system prompt for ChatGPT is me trying to solve this problem by telling it to challenge me, that I value the truth, and that I enjoy being told that I'm wrong.
Some things still slip through, but overall it's helped attenuate those kinds of behaviours
so, we’re now entering the polytheistic era - the time of many flawed and capricious gods, like the Greek or Norse pantheons. fair to assume that AGI will bring the one true God that sweeps them all away?
If you're interested in the dynamics of people perceiving AI to be conscious, check out Robert Saltzman's Substack ( https://robertsaltzman.substack.com/ ) and his book Understanding Claude ( https://www.amazon.com/dp/B0G5L9F998/ ). Saltzman has probed deeply into the "mind" of Claude and the minds of Claude users and offers compelling explanations.