* Its likely that the slowing of the tech job market wasn't caused by AI, but by a change in the tax code (Section 174) and higher interest rates (companies over-hired during the pandemic when funding was abundant).
* LLMs may or may not increase developer productivity [1], and they definitely cannot replace software engineers entirely (and I don't think they ever will - but it depends who you ask)
* Anecdotally, finding a summer internship wasn't easy for me, but it also wasn't any harder than it was for my peers in other programs (engineering, finance, etc.). Job hunting is a skill that I think many people in CS don't have because it used to be easy.
* I used an agentic IDE extensively to code for my on-campus research job. I still enjoyed the job a lot, and even as an rookie developer, I still felt I played a very valuable role in my job that LLMs could not replace.
[1] https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...
I think something like a medical doctor or a dentist is a much safer bet. They have basically always been able to maintain high salaries for their work.
Medical doctors are well oversupplied where I live. COVID hit them particularly hard and they were "drafted". A 35 year old doctor might be paid around the same as a 28 year old programmer.
Back when I graduated, software salaries were terrible, barely above minimum wage. Most quit, became teachers, bakers, fashion designers, etc. I quit to start a cafe, somehow got pulled back into software to raise funds to start a coffee franchise, and simply outlasted the other native Android developers.
Whatever seems oversupplied today will be undersupplied in 5-10 years, especially as advice as "don't do a degree" goes around.
So sure, if you're looking for high income and job security, and the work itself is irrelevant then that's one way to decide.
But clearly programming and doctoring take place in very different work environments, and are very different in many ways.
Personally I always enjoyed programming. Frankly, I don't like sick people, so I'd be a terrible doctor. I graduated at a time when Dr's were well paid, programming not so much. But the wheel turns, and I'm really pleased I chose the path I did.
40 years later, and I'm right where I'm supposed to be :)
As to AI, the current crop of AI (which is bubbling very well, and I believe is running itself quickly towards yet another AI winter [https://en.wikipedia.org/wiki/Ai_winter]), it may be useful for replacing some percentage of "head count" programmers, it is not likely to be replacing the actual creative, good, programmers (i.e., the one's often referred to as 10x). So the best advice I can give is to strive to really learn your craft such that you can function more towards the 10x side of the spectrum than the "head count" side. That will give you the best assurance of success at present.
With that being said, I did use my economics degree to get my foot in the door doing data analytics then I transitioned to SWE. You'll find you way, if being a SWE is truly your passion.
I graduated with a CS degree in 2012 so I fully benefited from the tech boom. If I were a senior in high school in 2025 knowing what I know now, I would probably go into Civil Engineering.
If you got into programming when you were a kid, can’t stay away from programming because you love it, and pursuing CS was an expression of this, then you should stick with it.
It used to be that there were only the second type because being a nerd wasn’t cool or lucrative. So this is just a return to baseline.
The people writing COBOL and FORTRAN on mainframes - I got my start writing C and FORTRAN on DEC VAX and Stratus VOS mainframes - didn’t speak about the joys of programming. They clocked in, clocked out and went on about their lives.
I doubt in 30 years across now 10 jobs I’ve met more than a handful of people that don’t have outside interests that they talk about at lunch outside of computers.
Whether he “loves” it or not is immaterial in the decision process. Whether someone will pay him so he can support his addiction to food and shelter is.
How many of the 2.7 million+ developers in the US wake up excited that they are creating Yet Another CRUD enterprise app ir some bespoke internal app that will never see the light of day outside of their company and thats the life of most developers world wide.
What I meant was that I think we're going through an extinction event for programmers of adaptability below a certain threshold. What will get a programmer through this period is going to be a drive and passion for the act of growing and a deep interest in the field. That's because change is here and the "jobs" are threatened.
One thing I've learned about AI coding is that you have to know what you want, and you need to ask for it. That's the condition where AI coding works. It does not do well with ambiguity and moonshots.
So ambiguity and moonshots are still the domain of the humans. Anyone with a repeatable job that's a slog is likely to find themselves on the sidelines soon if they don't adapt.
> "They clocked in, clocked out and went on about their lives."
Yeah, these folks are going to have a tough time. I don't wish it upon them, but that's where we're headed. And so if a person is just choosing programming because they want stability and a good job primarily, I say they should avoid it.
FORTRAN was my first language in the 60s and I ENJOYED using it until "better" languages came along.
I debugged COBOL and once taught SQL to COBOL programmers while refusing to write anything in it.
I had my best fun with mainframe Assembler and CMS Pipelines.
Hopefully you are getting a degree in order to educate yourself, broaden your skills and thinking, and have deep enough knowledge and critical thought to bring completely new ideas and innovations to your area of expertise.
With that goal in mind, CS as a degree absolutely gives you the base knowledge needed to roll with whatever changes come to the industry. To be one of the people who can sort through the hype to understand where AI truly works or does not, and build working solutions using whatever new tech comes to the world. If your program is fulfilling the goal of making you one of those people, it is worth sticking with it. If it is not making you one of those people, we're back to telling you just to go get a job.
But please do not fall into the trap of thinking that the purpose of a college degree is just to pass some gate to get your first job.
You can do extraordinarily well as a founder if you find an opportunity, get it to market and build a moat that competitors can't surmount.
Ideally, I'd first try out or explore the intended field of work one wants to be a part of before fully committing time, money, and energy on credentials for it. This might take the form of one or more internships, reverse interviews of people in that field, and/or finding interviews about a particular person in that role or company. For anyone already enrolled, I'd still check out internships and such but only change majors if multiple internships were decidedly terrible.
Follow fun for you that others find tedious or uninteresting where there is great value requiring human-in-the-loop expertise or effort, e.g., find a defensible niche not exposed to 100% automation. That might be a STEM or financial specialty or sub-specialty that you find more interesting like biomedical informatics, data science, statistics, accounting, or actuary because becoming a generic software engineer is as risky now as becoming a generic systems administrator.
Ultimately, it's a personal decision that cannot be offloaded to others requiring some experimental research/trial and error in the real world™. Plus, it needs some luck and finding a mostly positive working environment which tend to be in short supply.
Good luck.
https://mastodon.social/@UP8/115500560220694978
and they merged the Computer Science, Information Science and Data Science programs into one big department that enrolls 2000+ students
https://milestones.cis.cornell.edu/
So it is definitely something a lot of people are going into and a person who doesn't want to face a bubble pop might consider something else. One good thing at Cornell is that we have a data science minor that anybody can take. I went to a talk by an English professor for instance who applies quantitative methods and data visualization to literary criticism.
https://www.its.caltech.edu/~matilde/DarkBrightness.pdf
I've seen others argue for schizotypy using the same biographical examples rotfl.
My working slogan now is "Schizotypes transmute coffee to coin (value)."
The flip side: (functioning) Schizotypes (adjoining academia) have lost their obsession with magic, AND self-diagnosed this loss with depression/anxiety/senescence etc. it's more subtle imho, but CS is a confounding factor* (muddling the pipeline between "dopamine" and "coin")
As to the below.. it's why I try to keep my credibility inscrutable on this site :)
*Used to be Math if you follow the Erdos quip and understand theorems as a long-now stand-in for "Coin"
A bad choice if it is something you don't want to do.
And a reasonable choice if neither of those apply.
Good luck.
