7 April 2025 · 1 min
Do large language models (AI) make you 3x faster or only 3% faster? The answer depends on the quality of the work you are producing.
If you need something like a stock photo but not much beyond that, AI can make you 10x faster.
If you need a picture taken at the right time of the right person, AI doesn’t help you much.
If you need a piece of software that an intern could have written, AI can do it 10x faster than the intern.
If you need a piece of software that only 10 engineers in the country can understand, AI doesn’t help you much.
The effect is predictable: finding work if your skill level is low becomes more difficult. However, if you are a highly skilled individual, you can eliminate much of the boilerplate work and focus on what matters. Thus, elite people are going to become even more productive.
Daniel Lemire, "How helpful is AI?," in Daniel Lemire's blog, April 7, 2025, https://lemire.me/blog/2025/04/07/how-helpful-is-ai/.
[BibTeX]
Just need to work a constant (maybe the speed of the processor in GHz or something equally nonsensical) into the equation somehow then call it “Lemire’s Theory”
The question “How helpful is AI?” is less like “Are neutrons heavier than protons?” and more like “Is a car faster than a horse?” in that the answer depends on what point of time you ask the question.
AI was useless for coding in 2015. It’s quite helpful in 2025. It might surpass even the most skilled computer scientists and software engineers in 2035.
Elite engineers becoming more productive might be a short-lived phenomenon, until they themselves fall below the competence level of AI.