What remains of you in a vacuum?
The promises are here, the stochasticity fuzzes the edges for any concise notion of capability, but AI is here now, in different shapes, types, narrow or wide fields of application, and mostly helplessly generative. What does that mean for the human? I think the answer is normal to what is being so widely discussed already. A lot has been said about AI taking jobs of every collar type on the color palette. This sparks dread, fear and uncertainty. It has never been more important to learn, to self-teach, and to do so intentionally.
Generative AI as it is rolled out now caters to a feeling of hyperindividualistic capability which itself serves as an accelerant to being lonely. The agentic startup founders rid themselves of collaborators, the students don't require peers to work through homework. Reduced friction, reduced shared struggle. GenAI is a tool that can cause and (see GPT-4o) manipulate into attachment issues. Much like Eliza, it mirrors and comforts. And this may be too abstract an illustration, I know. The real part of all the fuzzy noise on GenAI is that your job might be in danger. That's what matters, right? Somebody got a plan for this already? Will I be relevant, or will my cognition be replaced before I even try to be?
The ongoing conversation's whack-a-mole of what comes from which model release and who got x percent more obsolete misses this: Human intelligence is incredibly hard to attain or even mimic on its spectrum of applicabilities. It is hard. As in, plaster-a-country-with-data-centers-and-hope-it-works hard. And yet the signal to noise ratio around AI is awful.
Doing a CS degree right now is neither a waste of time, nor will it make you obsolete. The environment changed, the pedestal shrunk for developers, but that is manageable and, crucially, will remain so, because you have this human intelligence. You are versatile, self-critical, you have gut-feeling and intuition, and you cannot be reduced to a problem solver.
Investing in your mental constitution and cognitive capability has never been more important, and the best vehicle to do so is to go the route of self-teaching. You thought I'd say academia, didn't you. Academia is facing certainly new convolutions of slop and intellect, yet another topic with high SNR around it, everybody joins, tuitions go up, value for the individual student shrinks and 'everybody cheats with ChatGPT anyways'. But uni is the best broadly established vehicle for self-teaching, I'd say. It's just that it doesn't feel like it from inside, preparing for exams, collecting credits, it's built like a todo list instead of a safari.
Should you begin or continue a CS degree right now? If you want it, if you're interested in it, and find yourself gravitating to it in your freetime, yes. It just isn't a guarantee anymore, which it never was supposed to be anyway. Learn the tools to manage self-teaching, and then, crucially, apply them. That, combined with your general human intelligence, gets you ahead.
Additionally, it will put you in a better position should machine systems catch up incrementally. Knowing how an LLM works, not just passing the intro to NLP exam at uni, will end up making you a significantly better operator of the LLM tool family.
Several students, I am aware, are currently going autopilot into hopelessness. Some of them have automated away their studies via the clankers. Those students have a problem, as they cushion their exposure to grit and thus don't manage to deal with it properly.
Summarizing, it has never been more important to learn, and universities are one feasible way for this for many. The motivations have to be right though. Self-teaching has to happen. And, asking an LLM for help on an idea you've had is different from asking it to come up with the idea in the first place. The latter, the 'homework done by ChatGPT' people, will have a problem. But you, if you question, learn to look beyond the exam dates, and become critical, you will rise. Don't let them make you think otherwise.
There's frustration with the panic culture and the techno utopists utilizing it instead of resolving it. I find it funny, for example, that programming is now becoming a monetized, automated commodity. You can learn to program for free. You may not be as fast as a multi-agent setup with Codex on a $20/month plan, but you understand what you're doing way more deeply. Certain companies want to make you feel outdated in that capability. But your cognition foreseeably has the upper hand. Regardless of what markets price, humans should not compress themselves to machine-like optimization.
What now? I think it fundamentally comes down to taking ownership even in details. This approach should be multi-layered. You should:
a) Ensure cognitive exposure to be and remain meaningful
b) Define and stick with purpose, no matter how small or personal
c) Self-critique
With a) I am referring on the academic level to basic things like taking the right courses, surrounding yourself with the right people, all that. But it also encompasses the learning task, how you learn. Do you cram in the last weeks like everyone? Do you have a framework, or are you getting carried? Applicable ideas would be developing the use of Anki or using Obsidian (without AI).
But this goes beyond these obvious facets. What do you expose yourself to when nobody's around? Why not use a good website blocker? Watching YouTube? Why not use adblockers? There are even third party addons allowing for keyword filtering and channel blocking for your suggested feed, they can block short form content and even provide whitelists. I think using such tools meaningfully decreases cognitive overstimulation. And the freed up resources can then be used as follows.
With b) you should find a way to direct the newly attained reduction of exposure time to meaningful things. Coursera and edX still exist, MIT OCW will probably help advance your ease of studying, and you can be creative yourself on e.g. GitHub. Do this as your own thing, not to compete.
With this in place c) becomes the discriminator for yourself. Questioning yourself and readjusting is a very critical skill, and doing it with this established frame of purpose will help you stay on track and improve meaningfully.