[brabel]

Can confirm. We don’t know if AI really is about to make programmers who write code by hand obsolete, but we sure as hell fear our competitors will ship features 10x faster than us. What is the logical next step?? Invest lots of money on AI or keep hoping it’s a fad and risk being left in the dust, even if you think that risk is fairly small?

[dgxyz]

Perhaps stop entering into saturated markets and using AI to try and shortcut your way to the moon?

There's no way any LLM code generator can replace a moderately complex system at this point and looking at the rate of progress this hasn't improved recently at all. Getting one to reason about a simple part of a business domain is still quite difficult.

[NitpickLawyer]

> and looking at the rate of progress this hasn't improved recently at all.

The rate of progress in the last 3 years has been over my expectations. The past year has been increasing a lot. The last 2 months has been insane. No idea how people can say "no improvement".

[coldtea]

>The past year has been increasing a lot. The last 2 months has been insane

I wonder is there are parallel realities. What I remember from the last year is a resounding yawn at the latest models landing, and even people being actively annoyed in e.g. ChatGPT 4.1 vs 4 for being nerfed. Same for 5, big fanfare, and not that excited reception. And same for Claude. And nothing special in the last 2 months either. Nobody considers Claude 4.6 some big improvement over 4.5.

Sorry for closing this comment early, I need to leave my car at the house and walk to the car wash.

[qnleigh]

Yeah not that long ago, there was concern that we had run out of training data and progress would stall. That did not happen at all.

[coldtea]

Sure it did.

[zozbot234]

"My car is in the driveway, but it's dirty and I need to get it washed. The car wash is 50 meters away, should I drive there or walk?"

[votepaunchy]

Gemini flash tells me to drive: “Unless you have a very long hose or you've invented a way to teleport the dirt off the chassis, you should probably drive. Taking the car ensures it actually gets cleaned, and you won't have to carry heavy buckets of soapy water back and forth across the street.”

[dgxyz]

Beep boop human thinking ... actually I never wash my car. They do it when they service it once every year!

[surgical_fire]

If your expectations were low, anything would have been over your expectations.

There was some improvement in terms of the ability of some models to understand and generate code. It's a bit more useful than it was 3 years ago.

I still think that any claims that it can operate at a human level are complete bullshit.

It can speed things up well in some contexts though.

[NitpickLawyer]

> It's a bit more useful than it was 3 years ago.

It's comments like these that make me not really want to interact with this topic anymore. There's no way that your comment can be taken seriously. It's 99.9% a troll comment, or simply delusional. 3 years ago the model (gpt3.5, the only one out there basically) was not able to output correct code at all. It looked like python if you squinted, but it made no sense. To compare that to what we have today and say "a bit more useful" is not a serious comment. Cannot be a serious comment.

[munksbeer]

> It's comments like these that make me not really want to interact with this topic anymore.

It's a religious war at this point. People who hate AI are not going to admit anything until they have no choice.

And given the progress in the last few months, I think we're a few years away from nearly every developer using coding agents, kicking and screaming in some cases, or just leaving the industry in others.

[surgical_fire]

This is such a weird framing.

My comment was that I think AI is useful. I use it on a daily basis, and have been for quite a while. I actually pay for a Chat GPT account, and I also have access to Claude and Gemini at work.

That you frame my comment as "people who hate AI" and calls ir "a religious war" honestly says more about you than me.

It seems that if you don't think that AI is the second coming of Christ, you hate it.

[surgical_fire]

/shrug

I have no intention of changing your mind. I don't think of the people I reply to highly enough to believe they can change their minds.

I reply to these comments for other people to read. Think of it as me adding ky point of view for neutral readers.

Either way, I could use AI for some coding tasks back in GPT 3.5 days. It was unreliable, but not completely useless (far from it in fact)

Nowadays it is a little more reliable, and it can do more complex coding tasks with less detailed prompts. AI now can handle a larger context, and the "thinking" steps it adds to itself while generating output were a nice trick to improve its capabilities.

While it makes me more productive on certain tasks, it is the sort of the improvements I expected in 3 years of it being a massive money black hole. Anything less would actually be embarrassing all things considered.

Perhaps if your job is just writing code day in an out you would find it more useful than I do? As a software engineer I do quite a bit more than that, even if coding is the bit of work I used to enjoy the most.

[sweetheart]

The recent developments of only the last 3 months have been staggering. I think you should challenge your beliefs on this a little bit. I don't say that as an AI fanboy (if those exist), it's just really, really noticeable how much progress has been made in doing more complex SWE work, especially if you just ask the LLM to implement some basic custom harness engineering.

[coldtea]

>The recent developments of only the last 3 months have been staggering.

What developments have been "staggering"? Claude 4.6 vs 4.5? ChatGPT 5.2 vs 5? The Gemini update?

Only the hype has been staggering, and bs non-stories like the "AI agents conspire and invent their own religion".

[dgxyz]

I'll let you know in 12 months when we have been using it for long enough to have another abortion for me to clean up.

[AstroBen]

Why is it an all or nothing decision?

Do a small test: if you're 10x faster then keep going. If not, shelve it for a while and maybe try again later

[zibzob]

It's not possible to tell if you're 10x faster, or even faster at all, over any non-trivial amount of time. When not using a coding agent, you make different decisions and get the task done differently, at a different level of architecture, with a different understanding of the code.

[AstroBen]

You don't think it's possible you'd notice if you, or the people around you, produced 6 months of work in 3 weeks?

The case where it's not obvious is when the effect is <1.5x. I think that's clearly where we're at

[zibzob]

I think it's a lot harder than it sounds. First, nobody can estimate time well enough to know how long something would have taken without AI. And then, it's comparing apples to oranges - there's the set of things people did using an AI agent, and the set of things they would have done if they hadn't used AI, and the two are just not directly comparable. The AI agent set would definitely have more lines of code, that's all I could really say. Maybe it would also contain a larger maintenance burden, lower useful knowledge in humans, projects that didn't need to be done or could have been done in a smarter way, etc.

[coldtea]

This sounds like lowering expectations with extra steps!

[brabel]

It is an all or nothing decision. We were letting devs make a decision about AI use and it turned out just a few were using it. The rest were using it just a little. But we noticed people were still spending time on stuff that AI can easily do for them!! I know because I had to do stuff for devs who were just not able to do things quickly enough and I did it with AI in minutes. Quite disappointing. But my conclusion is that we need to push AI quite strongly and invest in the best services otherwise people don’t even try it, and when they do it if it’s some cheap service they quickly dismiss it. So here we are.