This content is locked. Please login or become a member.
Intentional Stance
I think one of my best ideas is the intentional stance, the idea that we automatically adopt the strategy of treating anything complicated and interesting, if we can, as an agent. What does it want? What does it know? What does it believe? It works, of course, when the agent is another person, a lion that might want to eat us, or a fish that might want to escape from us and our fishing rod.
Well, that’s fine as long as we’re living in a world of concrete objects. But what’s now happening is that AI is filling the digital world with fake minds, fake people that we are almost irresistibly drawn to treat as if they were real. I fear that they will inundate us with counterfeit people, and these will be very dangerous memes indeed because they will replicate, they will evolve, and we won’t be able to take our attention away from them.
Prompting for Truth, Not Truthiness
We use the intentional stance all the time to endow beliefs and desires and do the rational thing given those beliefs and desires. Current AI, large language models, so-called, things like ChatGPT and GPT-4, their goal is truthiness, not truth. They are more like fiction writers, historical fiction writers than historians. If it sounds right, if the words hang together, that’s the only test that has to be passed by what they output.
We’re different in that we take truth seriously. LLMs, AIs don’t yet really have the capacity to lie on purpose. They do formulate falsehoods that they then spread, and there’s only a difference of degree between us and them. We’re not used to having our technology in a position to ignore the truth and just feed us what makes sense to them. We want smart machines, not artificial colleagues. We don’t want them to be autonomous. We want them to be our epistemic slaves, to do our bidding, to tell us the truth, not to tell us what they think we want to know, but what we want to know.