Rather than program an AI with what we want it to know, create ones capable of learning and have users teach them. Experience is the best teacher. Make AI charming so that users will be motivated to put effort into teaching them.
Book: Exhalation: Stories
This is from Ted Chiang's novella The Lifecycle of Software Objects, which is included in the collection Exhalation: Stories.
The story follows Blue Gamma, an AI company that sells digital AI pets called digients. The company's philosophy of AI design is that experience is the best teacher. Instead of trying to program an AI with what we want it to know, sell ones that can learn and have the customers teach them. To motivate customers to put effort into teaching the digients, the company ensures that every aspect of the AI pets is charming, including their personalities, which the Blue Gamma's developers are working on.
Rereading this piece, which was published in 2010, makes me reflect on ChatGPT. Looking at ChatGPT, its interaction follows a similar philosophy, with shareability playing an essential role in incentivizing users to engage with it. By sharing screenshots of the conversation with ChatGPT on social media, we get more users playing with it, which helps improve the AI's reinforcement learning.
We probably don't have to wait too long to see AI pet startups pop up and adopt this same "philosophy".