One of the biggest problems with today’s AI models is that they tend to simply make up answers when they don’t know what’s going on, something called hallucinations.
You would think that the number of hallucinations would decrease over time, but according to internal tests from Open AI, the opposite is true. The o3 and o4-mini reasoning AI models produce more hallucinations than their predecessors o1, o1-mini, and o3-mini, Techcrunch reports.
In one of the tests, the o3 model hallucinated in 33% of responses, compared to 16% for the o1 and 14.8% for the 03-mini.
