Member-only story
ChatGPT and the flat Earth, some thoughts on “artificial intelligence”.
Imagine for a moment that ChatGPT was born in the 17th Century and that you could have asked it if the Earth is flat.
The answer would have been a supporting and unconditional yes. Galileo may have already published his thesis and defended that the Earth rotates around the sun, but his opinion would have been diluted in a tide of hundreds of pages written by scholars and proving without a doubt (no need here to change the temperature of ChatGPT), that the Earth is flat. Even God was on their side.
The ChatGPT at the time would have worked very similarly to ours. It would have selected the answer with the obvious, most represented, but wrong answer, the answer with the highest probability, or the average answer in a normal distribution curve.
The right answer was the answer of the outlier. Are we ready to more and more eliminate dissonant voices?
