Computerphile
How AI works
What can I learn?
University of Nottingham's 800-video archive featuring Rob Miles' acclaimed AI-safety and LLM explainers in 10–20 minute chunks. The best English-language source for going one level deeper than 'how does AI work' to 'what are its failure modes?'
Core insight
The most dangerous AI risks are not science fiction — they are technical properties of current systems (misalignment, deception, hallucination) that are well-understood by researchers but poorly communicated to the public.
How to use it in daily work
Builds technically grounded understanding of AI limitations and failure modes, making your advice more precise.
- After watching Miles' videos on AI alignment, you can explain the difference between 'AI doing what it was told' and 'AI doing what we actually wanted' — a crucial distinction for clients trusting AI with important tasks.