GPT-3 AI Successfully Mimics Philosopher Daniel Dennett: Can You Tell the Difference?

GPT-3 AI successfully mimics philosopher Daniel Dennett

AI Philosophy

The AI model has been trained by using Dennett’s answers to a variety of questions, including whether animals feel pain or even the favorite quotes of other philosophers. Researchers asked people from different groups to compare AI responses with Dennett’s actual answers, and then see if there was a difference. The researchers used the responses of 302 random online users who clicked on a link in Schwitzgebel’s blog, as well as 98 college graduates confirmed by Prolific’s online research platform, and 25 Dennett experts. Dennett’s work and philosophy were so well-known that it didn’t stop anyone from having trouble identifying the source.

Participants in the research platform only answered 1.2 questions correctly on average. Blog readers and experts both answered 10 questions. The average score for the blog readers was 4.8 out 10. The Dennett experts were not all 100% accurate. Only one expert answered nine questions correctly, and the average score was 5.1 out 10, just barely above that of the blog readers. The question that confused Dennett experts the most was about AI sentience. Specifically, if \”people could ever build a robot with beliefs?\”

Source:

GPT-3 AI Successfully Mimics Philosopher Daniel Dennett