Ask HN: SWEs how do you future-proof your career in light of LLMs?

As I recall, the biggest problem from the last go-around was the models and implementation were two different sets of artifacts and therefore were guaranteed to diverge. If we move to a modern incarnation where the AI is generating the implementation from the models and humans are no longer doing that task, then it may work as the models will now be the only existing set of artifacts.

But I was definitely in camp #2 - the in-house business applications. I'd love to hear the experiences from those in camp #1. To your point, once IBM got involved it all went south. There was a place I was working for in the early 90s that really turned me off against anything "enterprise" from IBM. I had yet to learn that would apply to pretty much every vendor! :)

I think fundamentally if all you do is glue together popular OSS libraries in well understood way, then yes. You may be replaced. But really you probably could be replaced by a Wordpress plugin at that point.

The moment you have some weird library that 4 people in the world know (which happens more than you’d expect) or hell even something without a lot of OSS code what exactly is an LLM going to do? How is it supposed to predict code that’s not derived from its training set?

My experience thus far is that it starts hallucinating and it’s not really gotten any better at it.

I’ll continue using it to generate sed and awk commands, but I’ve yet to find a way to make my life easier with the “hard bits” I want help with.

We recently had a few submissions about this topic. Most recently Ilyas talk. Further improvement will be a research type problem. This trend was clear for a while already, but is reaching the mainstream now. The billions of dollar spend goes into scaling existing technology. If it doesn't scale anymore and becomes a resarch problem again, rational companies will not continue to invest in this area (at least without the usual research arrangements).

> CPU-only custom 2D pixel blitter engine I wrote to make 2D games in styles practically impossible with modern GPU-based texture rendering engines

I’m curious what’s so special about that blitting?

BTW, pixel shaders in D3D11 can receive screen-space pixel coordinates in SV_Position semantic. The pixel shader can cast .xy slice of that value from float2 to int2 (truncating towards 0), offset the int2 vector to be relative to the top-left of the sprite, then pass the integers into Texture2D.Load method.

Unlike the more commonly used Texture2D.Sample, Texture2D.Load method delivers a single texel as stored in the texture i.e. no filtering, sampling or interpolations. The texel is identified by integer coordinates, as opposed to UV floats for the Sample method.

I'm 15 years in, so a little behind you, but this is also some observations from the perspective of a student during the Post-Dot-Com bust.

A great parallel of today's LLMs was the Outsourcing mania from 20 years ago. It was worse than AGI because actual living breathing thinking people would write your code. After the Dot-Bomb implosion, a bunch of companies turned to outsourcing as a way to skirt costs for expensive US programmers. In their mind, a manager can produce a spec that was sent to an oversees team to implement. A "Prompt" if you will. But as time wore on, the hype wore off with every broken and spaghettified app. Businesses were stung back into hiring back programmers, but not before destroying a whole pipeline of CS graduates for many years. It fueled a surge in demand in programmers against a small supply that didn't abate until the latter half of the 2010s.

Like most things in life, a little outsourcing never hurt anybody but a lot can kill your company.

> My prediction is that junior to mid level software engineering will disappear

Agree with some qualifications. I think LLMs will follow a similar disillusionment as outsourcing, but not before decimating the profession in both headcount and senior experience. The pipeline of Undergrad->Intern/Jr->Mid->Sr development experience will stop, creating even more demand for the existing (and now dwindling) senior talent. If you can rough it for the next few years the employee pool will be smaller and businesses will ask wHeRe dId aLl tHe pRoGrAmMeRs gO?! just like last time. We're going to lose entire classes of CS graduates for years before companies reverse course, and then it will take several more years to steward another generation of CS grads through the curriculum.

AI companies sucking up all the funding out of the room isn't helping with the pipeline either.

In the end it'll be nearly a decade before the industry recovers its ability to create new programmers.

> So, fellow software engineers, how do you future-proof your career in light of, the inevitable, LLM take over?

Funnily enough, probably start a business or that cool project you've had in the back of your mind. Now is the time to keep your skills sharp. LLMs are good enough to help with some of those rote tasks as long as you are diligent.

I think LLMs will fit into future tooling as souped-up Language Servers and be another tool in our belt. I also foresee a whole field of predictive BI tools that lean on LLMs hallucinating plausible futures that can be prompted with (for example) future newspaper headlines. There's also tons of technical/algorithmic domains ruled by Heuristics that could possibly be improved by the tech behind LLMs. Imagine a compiler that understands your code and applies more weight on some heuristics and/or optimizations. In short, keeping up with the tools will be useful long after the hype train derails.

People skills are perennially useful. It's often forgotten that programming is two domains; the problem domain and the computation domain. Two people in each domain can build Mechanical Sympathy that blurs the boundaries between the two. However the current state of LLMs does not have this expertise, so the LLM user must grasp both the technical and problem domains to properly vet what the LLMs return from a prompt.

Also keep yourself alive, even if that means leaving the profession for something else for the time being. The Software Engineer Crisis is over 50 years old at this point, and LLMs don't appear to be the Silver Bullet.

tl;dr: Businesses saw the early 2000s and said "More please, but with AI!" Stick it out in "The Suck" for the next couple of years until businesses start demanding people again. AI can be cool and useful if you keep your head firmly on your shoulders.