LLMs for coding are not even close to imperfect, yet, but the saturation curves are not flattening out; not by a long shot. We are living in a moment and we need to come to terms with it as the work continues to develop; and, we need to adapt and quickly in order to better understand what our place will become as this nascent tech continues its meteoric trajectory toward an entirely new world.

Instead, we have a tiny handful of one-off events that were laboriously tuned and tweaked and massaged over extended periods of time, and a flood of slop in the form of broken patches, bloated and misleading issues, and nonsense bug bounty attempts.

Then the people who congratulate the AI for helping get yelled at by the other category.

We'd still have more than tortured, isolated, one-offs. We should have at least one well-known codebase maintained through the power of Silicon Valley's top silicon-based minds.

If the future didnt turn out to be revolutionary, you now have done some "unnecessary" work at worst, but might've acquired some skills or value at least. In the case of most well off programmers, i suspect buying assets/investments which can afford them at least a reasonable lifestyle is likely too.

So the default position of being stationary, and assuming the world continues the way it has been, is not such a good idea. One should always assume the worst possible outcome, and plan for that.

Maybe if you work e-commerce or in the military.

But how do you even translate this line of thought for today?

Is you EMP defenses up to speed?

Are you studying russian and chinese while selling kidneys in order to afford your retirement home on Mars?

My point being, you can never plan for every worst outcome. In reality you would have a secondary data center, backups and a working recovery routine.

None of which matters if you use autocomplete or not.

I don't need to convince anyone that LLMs are enabling me to do a lot more. This is what makes this hype different. It has bones. Once you've found a way to leverage it, it's undeniably helpful regardless of your prior disposition. Everyone else can say they're not useful and it rings hollow because it obviously is to me. And thus probably useful to everyone else too.

They can’t write me a safety-critical video player meeting the spec with full test coverage using a proprietary signal that my customer would accept.

If you always say that every new fad is just hype, then you'll even be right 99.9% of the time. But if you want to be more valuable than a rock (https://www.astralcodexten.com/p/heuristics-that-almost-alwa...), then you need to dig into the object-level facts and form an opinion.

In my opinion, AI has a much higher likelihood of changing everything very quickly than crypto or similar technologies ever did.

If you want to convince skeptics talk about examples, vibe code a successful business, show off your success with using AI. Telling people it's the future and if you disagree you have your head in the sand, is wholly unconvincing.

Don't people and companies using AI lazily to put out low quality content blind you to its potential as well as the reality of what it can do right now. Look at Google's VO3, most people in the world right now won't be able to tell you that it's AI generated and not real.

This doesn't feel like that. The applications of generative AI have become self-evident to anyone that's followed their rise. Specific applications of AI resemble snake oil, and there are hucksters who pivoted from crypto to AI, but the ratio of legit use cases to scams isn't even close.

If anything, the incentives for embellishment have flipped since crypto. VC-funded AI companies will dreamily fire press releases about AI taking us to Mars, but it doesn't have the pseudo-grassroots quality of cryptocurrency hype. The average worker is incentivized to be an AI skeptic. The rise of generative AI threatens workers in several fields today, and has already negatively impacted copywriters and freelance artists. I absolutely understand why people in those fields would respond by calling AI use unethical and criticize the shortcomings of today's models.

We'll see what the next few years hold. But personally, I foresee AI integration ramping up. Even if the models themselves completely stagnate from this point on, there's a lot of missing glue between the models and the real world.

There's no guarantee a technology will take off, even if it's really, really good. Because we don't decide if that tech takes off - the lawyers do. And they might not care, or they might decide billing more hours is better, actually.

History majors everywhere are weeping.

The guiding principle of biglaw.

Attorneys have the bar to protect them from technology they don’t want. They’ve done it many times before, and they’ll do it again. They are starting to entertain LLMs, but not in a way that would affect their billable hours.

Look, we see the forest. We are just not impressed by it.

Having unlimited chaos monkeys at will is not revolutionizing anything.

When I graduated high school, I had never been or knew anyone who had ever been on the internet at all. The internet was this vague "information superhighway" that I didn't know really what to make of.

If you are of a certain age though you would think a pointless update to react was all the change ever coming.

That time is over and we are back to reality.

Or maybe they just know the nitty-gritty inherent limitations of technology better than you.

(inb4: "LLMs can't have limitations! Wait a few years and they will solve literally every possible problem!")

Software isn't like this. No one cares why you wrote the code in your PR. They only care about whether it's right.

This is why LLMs could be useful in one industry and a lot less useful in another.

I am not a software engineer but I just can't imagine my job is not automated in 10 years or less.

10 years is about the time between King – Man + Woman = Queen and now.

I think what is being highly underestimated is the false sense of security people feel because the jobs they interface with are also not automated, yet.

It is not hard to picture the network of automation that once one role is automated, connected roles to that role become easier to automate. So on and so on while the models keep getting stronger at the same time.

I expect we will have a recession at some point and the jobs lost are gone forever.

I've seen Claude and ChatGPT happily hallucinate whole APIs for D3 on multiple occasions, which should be really well represented in the training sets.

Recently I converted all the (Google Docs) documentation of a project to markdown files and added those to the workspace. It now indexes it with RAG and can easily find relevant bits of documentation, especially in agent mode.

It really stresses the importance of getting your documentation and processes in order as well as making sure the tasks at hand are well-specified. It soon might be the main thing that requires human input or action.

In fact, I built an entirely headless coding agent for that reason: you put tasks in, you get PRs out, and you get journals of each run for debugging but it discourages micro-management so you stay in planning/documenting/architecting.

With many existing systems, you can pull documentation into context pretty quickly to prevent the hallucination of APIs. In the near future it's obvious how that could be done automatically. I put my engine on the ground, ran it and it didn't even go anywhere; Ford will never beat horses.

Which means it's back to being a very useful tool, but not the earth-shattering disruptor we hoped (or worried) it would be.

Yet?

Fun to consider but that much uncertainty isn't worth much.

For other stuff this is obviously harder.

o3 came out just one month ago. Have you been using it? Subjectively, the gap between o3 and everything before it feels like the biggest gap I've seen since ChatGPT originally came out.

Do you want to be a jobless weaver, or an engineer building mechanical looms for a higher pay than the weaver got?

You could even go back to punch cards if you want to. Literally nobody forcing you to not use it for your own fun.

But LLMs are a multiplier in many mundane tasks (I'd say about 80+% of software development for businesses), so not using them is like fighting against using a computer because you like writing by hand.

Happy to hate myself but earn OK money for OK hours.

Tools and systems which increase productivity famously always put everyone out of a job, which is why after a couple centuries of industrial revolution we're all unemployed.

An LLM is more like outsourcing to a consultancy. Results may vary.

Depends on the context. You have to keep in mind: it is not a goal of our society or economic system to provide you with a stable, rewarding job. In fact, the incentives are to take that away from you ASAP.

Before software engineers go celebrate this tech, they need realize they're going to end up like rust-belt factory workers the day after the plant closed. They're not special, and society won't be any kinder to them.

> ...and even wiser to be in charge of and operate the replacement.

You'll likely only get to do that if your boss doesn't know about it.

Don’t let cynics rule your country. Go vote. There’s no rule that things have to stay awful.

We seem to agree as this is more or less exactly the my point. Striving to keep the status quo is a futile path. Eventually things change. Be ready. The best advice I've ever got work (and maybe even life) wise is to always have alternatives. If you don't have alternatives, you literally have no choice.

Those alternatives are going to be worse for you, because if they weren't, why didn't you switch already? And if a flood of your peers are pursing alternatives at the same time, you'll probably experience an even poorer outcome than you expected (e.g. everyone getting laid off and trying to make ends meet driving for Uber at the same time). Then, AI is really properly understood as a "fuck the white-collar middle-class" tech, and it's probably going to fuck up your backup plans at about the same time as it fucks up your status quo.

You're also describing a highly individualistic strategy, for someone acting on his own. At this point, the correct strategy is probably collective action, which can at least delay and control the change to something more manageable. But software engineers have been too "special snowflake" about themselves to have laid the groundwork for that, and are acutely vulnerable.

I do concur it is an individualistic strategy, and as you mentioned unionization might have helped. But, then again it might not. Developers are partially unionized where I live, and I'm not so sure it's going to help. It might absorb some of the impact. Let's see in a couple of years.

People have families to feed and lifestyles to maintain, anything that's not equivalent will introduce hardship. And "different" most likely means worse, when it comes to compensation. Even a successful career change usually means restarting at the bottom of the ladder.

And what's that "something else," exactly? You need to consider that may be disrupted at the same time you're planning on seeking it, or fierce competition from your peers makes it unobtainable to you.

Assuming there are alternatives waiting for you when you'll need them is its own kind of complacency.

> It might be selling all your stuff and live on an island in the sun for all I know.

Yeah, people with the "fuck-you" money to do that will probably be fine. Most people don't have that, though.

Hardship or not, restarting from the bottom of the ladder or not, betting on status quo is a loosing game at the moment. Software development is being disrupted, I would expect developers to produce 2-4x more now than two years ago. However, that is the pure dev work. The architecture, engineering, requirements, specification etc parts will likely see another trajectory. Much due to the raise of automation in dev and other parts of the company. The flip side is that the raise of non-dev automation is coming, with the possibility of automating other tasks, in turn making engineers (maybe not devs though) vital to the companies process change.

Another, semi related, thought is that software development has automated away millions of jobs and it’s just developers time to be on the other end of the stick.

"AI" is the latest iteration of snake oil that is foisted upon us by management. The problem is not "AI" per se, but the amount of of friction and productivity loss that comes with it.

Most of the productivity loss comes from being forced to engage with it and push back against that nonsense. One has to learn the hype language, debunk it, etc.

Why do you think IT has gotten better? Amazon had a better and faster website with far better search and products 20 years ago. No amount of "AI" will fix that.

The sentiment I often see online is that "fewer rules = more freedom". And as I grew more experienced, I find that the opposite is true. In a lot of cases, freedom is upheld by the rules themselves. And this is not an orwellian "war is peace" kind of thing, but rather a GNU Free Software kind of thing, where the perpetuity of the freedom is guaranteed with the rules of participation.

I think that this discussion went in the same direction with the unions. But, of course, freedom in itself can have different interpretations for different people.

I hate AI code assistants, not because they suck, but because they work. The writing is on the wall.

If we aren't working on our own replacements, we'll be the ones replaced by somebody else's vibe code, and we have no labor unions that could plausibly fight back against this.

So become a Vibe Coder and keep working, or take the "prudent" approach you mention - and become unemployed.

This "vibe coding" seems just another way to say that people spend more time refining the output of these tools over and over again that what they would normally code.

But there's going to be an inflection point - soon - as things continue to improve. The industry is going to change rapidly.

Now is the time to either get ready for that - by being ahead of the curve, at least by being familiar with the tooling - or switch careers and cede your job to somebody who will play ball.

I don't like any of this, but I see it as inevitable.

Clearly they do work in a general sense. People who don't want to code are making things that work this way right now!

This isn't yet replacing me, but I'm certain it will relatively soon be the standard for how software is developed.

Personally I’m thrilled that I can get trivial, one-off programs developed for a few cents and the cost of a clear written description of the problem. Engaging internal developers or consulting developers to do anything at all is a horrible experience. I would waste weeks on politics, get no guarantees, and waste thousands of dollars and still hear nonsense like, “you want a form input added to a web page? Aw shucks, that’s going to take at least another month” or “we expect to spend a few days a month maintaining a completely static code base” from some clown billing me $200/hr.

Using it to prototype some low level controllers today, as a matter of fact!

Any skilled developer with a decade of experience can write prompts that return back precisely what we wanted almost every single time. I do it all day long. "Claude 4" rarely messes up.

I have a hard time imagining an LLM being able to do arbitrary things. It always feels like LLMs can do lots of the easy stuff, but if they can't do everything you need the skilled engineer anyway, who'd knock the easy things out in a week anyway.

Wrote me:

- a SPI deserializer that sets a bit after 12 bits read in, to trigger a prefetch

- an SDC constraints file for the deserializer that correctly identified the SPI clock and bus clock as separate domains requiring their own statement

- a test bench that validated both that the prefetch bit was being set, and that it was being set at the proper time relative to the SPI clock

- a makefile with commands for build, headless test, and debug by loading the VCD into a waveform viewer

It always feels like LLMs can do lots of the easy stuff, but if they can't do everything you need the skilled engineer anyway, who'd knock the easy things out in a week anyway.

Nearly every part of the tool flow I just described, I would consider “tricky to get right”. Been doing this for ~15 years and it’s still tough to bootstrap something like this from scratch. ChatGPT-4o did this for me from zero in about 15 minutes.

I won’t lie: I love it. I can focus on the actual, bigger problems at hand, and not the tricky little details of HDLs.

People are either deluding themselves or ignorant of the capabilities of frontier models if they don’t believe LLMs offer a speedup in workflow.

I personally believe that most of the doubt and cynicism is due to:

1) a pretty big collective identity crisis among software professionals, and

2) a suspicion that LLMs make it so that anyone who is good at articulating the problem precisely no longer needs a software engineer as a translation specialist from specs to code.

I say this as an EE of ~15 years who’s always been able to articulate what I want, specifically, to a firmware counterpart, who then writes the code I need. I can turn years of practice in this skill into great prompts for an LLM, which effectively cuts out the middleman.

I really like it. It’s helped me take on a lot of projects that are just outside of my innate level of capability. It’s also helped me learn a lot of new things about these new software adjacent areas. ChatGPT is a great tutor!

> 1) a pretty big collective identity crisis among software professionals, and

> 2) a suspicion that LLMs make it so that anyone who is good at articulating the problem precisely no longer needs a software engineer as a translation specialist from specs to code.

... But "articulating the problem precisely" is a huge part of what software engineers do, and there's a mountain of evidence that other people are not very good at that.

Why would I add a subpar translation layer into the process of achieving my goals? There’s no inherent value in that.

Because you don't have a choice. Your thoughts are not code.

If the FPGA is mostly doing simpler stuff with lots of boilerplate, I can see current LLMs offering a lot for someone who doesn't regularly write code for them; I guess that's similar to the current case for software.

Using them to set up the initial flow is not a bad idea either - I know coworkers who use them to write the early code for a new system or driver, where it seems to work pretty well (probably because that's a huge part of the training set - loads of tutorials out there)

LLMs output are unreliable and productivity is still not proven for an end to end engineering cycle.

Remember how blockchain was going to change the world? Web3? IoT? Etc etc.

I've been through enough of these cycles to understand that, while the AI gimmick is cool and all, we're probably at the local maximum. The reliability won't improve much from here (hallucinations etc), while the costs to run it will stay high. The final tombstone will be when the AI companies stop running at a loss and actually charge for the massive costs associated with running these models.

Personally, I'm more interested in the political angle. I can see that AI will be disruptive because there's a ton of money and possibly other political outcomes depending on it doing exactly that.

Have you tried talking to ChatGPT voice mode? It's mind blowing. You just have a conversation with it. In any language. About anything. The other day I wanted to know about the difference between cast iron and wrought iron, and it turned into a 10 or 15 minute conversation. That's maybe a good example of an "easy" topic for LLMs (lots of textbooks for it to memorize), but the world is full of easy topics that I know nothing about!

> but each succeeding iteration seems to be more disappointing

This is because the scaling hypothesis (more data and more compute = gains) is plateauing, because all text data is used and compute is reaching diminishing returns for some reason I’m not smart enough to say why, but it is.

So now we're seeing incremental core model advancements, variations and tuning in pre- and post training stages and a ton of applications (agents).

This is good imo. But obviously it’s not good for delusional valuations based exponential growth.