Software engineering knowledge is no more undermined by Gen AI than calculators undermined mathematicians. But there are very few left in the field of mathematics or statistics who insist on doing every calculation on paper or in their head. - ThreadSky

principalengineer.com • 30 days ago

Software engineering knowledge is no more undermined by Gen AI than calculators undermined mathematicians.

But there are very few left in the field of mathematics or statistics who insist on doing every calculation on paper or in their head.

Comments

timfish.dev•30 days ago

The analogy falls apart when you consider that calculators always give you the correct answer when the inputs are correct 😂

principalengineer.com•30 days ago

When are software requirement "inputs" ever likely to be correct, whether expressed to a human or computer?

timfish.dev•30 days ago

Yes, the analogy also falls apart there.

timo.mihaljov.com•30 days ago

I don't think they are very analogous technologies. A calculator automates a completely mechanical process that always produces the only correct answer, whereas generative AI replaces a fuzzy process during which we constantly recalibrate our conception of what correct even looks like.

timo.mihaljov.com•30 days ago

Although the above essay is about the use of generative AI in academia, if one subscribes to Naur's view of programming as theory building (as I do), its key points remain very relevant. In both cases the textual artifact is not as important as the experience and understanding behind it.

timo.mihaljov.com•30 days ago

(I don't expect you to read these articles, but I included the links just in case you or someone else reading this wants to dive deeper into the arguments or simply to understand what the hell I am on about 😄)

principalengineer.com•30 days ago

They are not analogous, because Gen AI uses fuzzy inputs to replace fuzzy processes with fuzzy outputs.

But the result for work is similar: they free us from some low-value work, so we can focus on more valuable work, that requires real intelligence.

timo.mihaljov.com•30 days ago

I think we agree about the immediate effects, and it's the second-order effects where our views diverge and I veer into skepticism. I suspect that much of what we call low-value work isn't really wasted effort but actually builds the mental muscle we need to do the high-value work effectively.

Comments

Posting Rules

Reply