How AI chat really works. Your chatbot isn't really thinking.

On two occasions I have been asked, — 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' ... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.
Charles Babbage. "Passages from the Life of a Philosopher" (1864)

I love using AI. I use it all the time. It is so useful. It both increases my efficiency and allows me to do more ambitious things. But I'm increasingly worried by its effect on people and society. Many people interact with ChatBots as if they are talking to a person, or at least a thinking entity, and trust the outputs far too much. Some people are even developing emotional relationships with ChatBots.

ChatBots like ChatGPT and Claude are based on Large Language Models (LLMs). In this session we will be looking at how LLMs are created from data, and how they work. We will also consider the question of whether LLMs actually think and reason, and why it's highly unlikely they are conscious or experience emotions.

Understanding LLMs will help you use them better, and I expect will make you a bit more skeptical of their responses.

This session will be led by James.

Bring wine/beer/drinks/food/snacks/whatever if you want to.

Our events fill up very quickly. To avoid missing out on future events, join the WhatsApp group via this link:
https://tally.so/r/w4G5L5

Press the buzzer marked Semi sotano 2 (bottom left on the panel) to be let in. It's down the stairs, first door on the right.

Image: Closer than we think. The future world according to the illustrator Arthur Radebaugh, 1959.