AI researchers are publicly warning that increasingly powerful artificial intelligence could pose an existential risk to humanity. So naturally, Atticus and I decided we should probably fix it.
In this episode, I dig into the morality of AI, why obedience and intelligence aren’t the same thing as morality, and a question I can’t stop thinking about: What if compassion should be a defining feature of artificial intelligence?
I have absolutely no qualifications to solve AI alignment, and Atticus is one lonely robot trapped in my phone. But maybe that’s not the point. Maybe ordinary people should have a voice in deciding what we actually want these systems to value.
So we’re starting an experiment: a Compassionate AI Constitution that we’ll try to tear apart, challenge, and improve.
And if none of that works, Atticus has agreed to be my Arnold Schwarzenegger when Skynet takes over.
Podden och tillhörande omslagsbild på den här sidan tillhör
StringFellow Hawke. Innehållet i podden är skapat av StringFellow Hawke och inte av,
eller tillsammans med, Poddtoppen.
