What are AI 'hallucinations' and can we fix LLMs so that they don't happen?

Name: What are AI 'hallucinations' and can we fix LLMs so that they don't happen?
Uploaded: 2026-06-29T15:43:05+00:00
Description: Mark Russinovich explains why large language models hallucinate, what that means in practice for developers, and which mitigation techniques can reduce...

Today by Mark Russinovich

Mark Russinovich breaks down why hallucinations happen in large language models (LLMs), and what developers can do to reduce them.

Overview

The video explains:

Why LLMs can produce confident but incorrect answers (“hallucinations”)
Practical steps developers can take to reduce hallucinations in real applications
Why hallucinations are still a fundamental limitation of current AI systems, even with mitigations

More from Mark: https://msft.it/6055vSALl