Unit D · AI and chatbots
What we are making: an accurate model of the machine. You have spent seventeen sessions refusing to trust anything you could not explain. Apply the same discipline here.
A chatbot is a program that predicts the next chunk of text, over and over, until it has produced an answer.
That is not a simplification for school. That is the mechanism, and everything surprising about these systems — the good and the bad — falls out of it.
In Unit C you wrote a classifier. You chose the threshold, you can point at the line, and you can say exactly why it decided what it decided. That is worth holding onto as a reference point.
Because for an enormous amount of ordinary text, the most likely continuation is the true one. Paris really does follow “the capital of France is” in almost everything ever written. Truth and likelihood line up most of the time.
The failures happen precisely where they come apart: things rarely written about, things written about after training stopped, and things where a fluent answer exists but a true one was never recorded.
You have spent the year asking one question about every device: under what conditions does this stop being reliable? Ask it here.
| Kind of question | How reliable | What to do |
|---|---|---|
| Common, well-documented facts | Very reliable | Still verify anything that matters |
| Explaining a concept | Excellent — its strongest use | Check it against one other source |
| Code in a popular language | Usually good, plausible when wrong | Compile it. Test it. Never paste unread |
| Anything local or recent | Unreliable, and does not say so | Treat every specific as unverified |
| Its own confidence | Not correlated with correctness | Confidence is not evidence |
In this session