Ask HN: Has generative AI's success made a CS degree more of a necessity now?

Generative AI cannot do the job of software devs yet. This is the truth. I do not know what will happen in next 5 years but at this stage generative AI cannot do software development. Sure it can write snippets of code which are helpful.

A CS degree does not necessarily teach you how to build complex software systems. So experienced devs are not replaceable by AI yet. But fresh CS grads with just some problem solving based coding skills can be replaced by AI. So I think it is important to learn how to build real systems end-to-end and showcase that experience.

Because only the highly skilled will be needed and the highly skilled will need CS degrees to distinguish themselves from the now-useless react devs?

Maybe...? But if that ends up being the case, we'll be on the other side of an economic event horizon and we'll all be so thick in the middle of who the hell knows what at that point that whether you have a degree or not, you better buckle the fuck up.

If you are referring to building ML services, why would a CS degree be necessary? The technology behind generative deep learning falls more in the domain of statistics and information theory than pure computer science (in the discrete mathematics and algorithms/data structures sense of the word). A foundation in scientific programming and a good intuition for multivariable calculus and Bayesian statistics goes much further. I would wager that the perfect candidates for ML research are found in physicists, electronic engineers and other quantitative disciplines (perhaps computational chemists too). Most programmers' CS training are nowhere near enough to understand for example autoencoders at the information theoretic level. How many enterprise java programmers do you think can explain what latent space is? Calling API endpoints and building products around prompts is not the same as actually understanding ML to be a practitioner.

Education will become more important but not necessarily a CS degree. AI may soon eliminate the beginner experts. How to write code will become largely unimportant and instead how to write applications will become supremely important. If you cannot apply simple organizational skills or communicate then your education isn’t very valuable in a practical sense.

Can't really see how the two would be related. I use ChatGPT and while it definitely makes you more productive, it is basically just a faster, more user-friendly way of tapping into community knowledge compared to stuff like Stack Overflow.

A lot of industry solutions will require more than simple prompt engineering or running the QuickStart from a tutorial. Understanding and grasping the fundamentals of these models will be a huge asset, and CS degrees could definitely help with that

You definitely need CS background to deeply understand what's going on under the hood, but, the capabilities of GPTs will also intimidate a lot of people avoid putting 4 years time on CS because most jobs can be done by AI (in the future), include most coding tasks.

So only those really love CS will choose CS major by no means, while others may just want to be a prompting engineer.

Who knows? Maybe a greater emphasis on "QA" (aka testing) again if they're going to actually use generative AI to produce code.

I have seen zero evidence that “AI’s” can take over the job of even junior developers.

It can’t really code, that’s overblown hype / PR.