My current hypothesis is that as AI forces software development down less and less deterministic pathways, I suspect that the value of a basic CS degree will diminish relative to humanities training. Comfort with ambiguity, an ability to construct a workable "theory of mind", and to construct unambiguous natural-language prompts will become more relevant than grokking standard algorithms.
The reverse most certainly is not true, and even if it were it wouldn't matter.
Humanities advocates have been hoping for the demise of valuable STEM degrees for at least the last 30 years. It's not happening for many reasons, of them being: All the skills you listed are also taught in an engineering and rigorous CS curriculum, plus those degrees provide validation that the individual is intelligent and determined enough to complete coursework that most people cannot.
I dunno, man. The difficulty (and resentment of having to even take them) most STEM majors had in my college-level writing classes causes me to doubt that, as does the general reaction on this board to any kind of problem / domain with irreducible ambiguity. But look, I'm not talking about the top ~10%, or whatever: the really smart kids can adapt to whatever gets thrown at them[0]. I'm doubtful that a 50th-percentile or below CS degree / student will retain the value that they've recently had - and given what I read on here about the present job market for new grads on here, that's maybe already happening.
Anyway, I had to pick one, my money'd be on philosophy degrees rising in value: they're already sought out by financial firms. Have you seen the sort of analytical / symbolic reasoning they do?
[0] In fact, in case you didn't know, rigorous humanities programs and research involve an awful lot of statistics and coding, even though the dinosaurs that run the MLA and most English departments aren't able to handle it.
> I dunno, man. The difficulty (and resentment of having to even take them) most STEM majors had in my college-level writing classes causes me to doubt that, as does the general reaction on this board to any kind of problem / domain with irreducible ambiguity.
I don't think most STEM majors would be outstanding English Literature (or whatever humanities program you prefer) majors, but I do think they could manage to obtain a degree. Very, very few humanities majors could get an engineering degree.
And yes, the writing classes they force engineers to take are largely pointless and not enjoyable. Everyone with a degree got through them though, and I have to imagine the percentage of STEM students who washed out on that and not organic chemistry, compiler design, differential equations, etc. is extremely small (it was 0 out of the hundreds of people I knew at my school).
> But look, I'm not talking about the top ~10%, or whatever: the really smart kids can adapt to whatever gets thrown at them[0].
Sure. Very few of these kids are going into publishing, because they'll have more lucrative options and will pursue them.
> I'm doubtful that a 50th-percentile or below CS degree / student will retain the value that they've recently had - and given what I read on here about the present job market for new grads on here, that's maybe already happening.
That may be, but they're still in better shape than a 50% percentile humanities degree holder, who also is having the value of their skillset eroded by AI.
> Anyway, I had to pick one, my money'd be on philosophy degrees rising in value: they're already sought out by financial firms. Have you seen the sort of analytical / symbolic reasoning they do?
Lol, they are not "sought out" in any sense of the word. Philosophy majors at top tier schools are sought out because everyone at the school is sought out, not because they majored in philosophy.
And yes, I took a number of philosophy classes in college as an undergrad because they were easy (have you seen the analytical/symbolic reasoning required of EE or CS majors? It's a lot more difficult that what is required of philosophy majors).
> [50th percentile CS grads] are still in better shape than a 50% percentile humanities degree holder, who also is having the value of their skillset eroded by AI.
That's the crux of it, and right now it appears to me that the ability to write unambiguous natural language prompts - in a variety of contexts, not specifically heavy-duty dev work - is going to be increasingly valuable. The 50th percentile english / philosophy grad is better at that than the 50th percentile CS major - while, at the same time, the bottom rungs of the developer ladder appear to have been kicked out.
I'm trying very hard not to make this into a "who's smarter?" question. That's a well-trodden and pointless argument, particularly if money is going to be the measuring stick. Besides, if that's where we're going, the finance bros and C-suite win, and do either of us think they're the geniuses in the room?
But, we'll see. We're living in Interesting Times.
> That's the crux of it, and right now it appears to me that the ability to write unambiguous natural language prompts - in a variety of contexts, not specifically heavy-duty dev work - is going to be increasingly valuable. The 50th percentile english / philosophy grad is better at that than the 50th percentile CS major - while, at the same time, the bottom rungs of the developer ladder appear to have been kicked out.
We don't agree here. I see no evidence that the average humanities major is better at writing unambiguous natural language, nor that it will be a partcularly valuable skill. Most people are incapable of understanding and describing a complex series of steps, including their side effects and tradeoffs regardless of the language used to describe them.
> I'm trying very hard not to make this into a "who's smarter?" question. That's a well-trodden and pointless argument, particularly if money is going to be the measuring stick. Besides, if that's where we're going, the finance bros and C-suite win, and do either of us think they're the geniuses in the room?
That's my point, there's no avoiding this. Standardized test scores used as part of college admissions are intelligence tests and income is highly correlated with intelligence. We have all these proxies that are providing the answer to this question.
And the hedge fund managers and CEOs of large companies are very intelligent on average (I'm sure some aren't but they are the outliers, not the other way around). Just like there are some very intelligent social workers, artists, and unemployed people, but the averages are what they are for various fields for a reason.
> I see no evidence that the average humanities major is better at writing unambiguous natural language
If you'd marked enough undergrad papers you would have. :-)
> Most people are incapable of understanding and describing a complex series of steps, including their side effects and tradeoffs regardless of the language used to describe them.
That's true!
But... The AI promise is that users won't have to do all of that part. They'll describe an end-state, and the machine will work out the steps needed to get there, asking clarifying questions along the way. If that's true, then skills like writing and interface design and "taste" and all the other "non-engineering" parts of making things rise in importance relative to the engineering skills that have been handed over to the machines.
That's a big "if", of course, and the machines aren't there yet, but that's what's promised. If it comes to pass, then I like my prediction (for, at least, the 50th percentile of both groups). If not, not.
I suggest you revisit your hypothesis with a little less bias.