Match the development era to the typical unit of work.

According to the MIT NANDA result presented in the lecture, what characterized the 5% of GenAI pilots that worked?

True or false: A benchmark score above about 80% necessarily means the benchmark remains a strong discriminator among models.

What conclusion does the lecture draw from the rapid growth of commits and pull requests?

How should an engineer treat a model's written chain of thought according to the lecture?

Which statement correctly distinguishes HumanEval from SWE-bench?

True or false: Felt productivity and measured productivity are the same quantity.

Why can a 27-billion-parameter model quantized to 4-bit run on a desktop-class machine?

Why is prompt injection especially relevant to coding agents?

Match each future scenario to its central claim.