Topic
2 posts on working with ai.
A confident wrong answer is produced by exactly the same process as a confident right one. The fix is not asking the model to be careful. It is making every claim carry its method.
Adding a second agent to check the first one buys you much less than it looks like. Agreement only means something when the methods are independent.
# Working With Ai
2 posts by Mihajlo Maiga.
## [An AI will not tell you it is guessing](https://mihajlomaiga.com/blog/an-ai-will-not-tell-you-it-is-guessing)
4 August 2026
A confident wrong answer is produced by exactly the same process as a confident right one. The fix is not asking the model to be careful. It is making every claim carry its method.
## [Two agents agreeing is one observation](https://mihajlomaiga.com/blog/two-agents-agreeing-is-one-observation)
4 August 2026
Adding a second agent to check the first one buys you much less than it looks like. Agreement only means something when the methods are independent.
[All writing](https://mihajlomaiga.com/blog)