>> telling it to refactor so-and-so class according to whatever principles

I've been revisiting code that I wrote a while back looking for things that I could carry forward into other projects if it were packaged up better. Convert this module into a class, break these major areas of functionality out into microservices, etc. AI is a whiz at that kind of thing and now I have a much larger set of reusable utilities.

Yes I find that the customer frequently doesn't know what is feasible to do, or is optimally desirable for his use case, or understand the trade-offs that might be involved.

What's nice about our current AI-assisted era is that you can quickly slap together a prototype or two and show it to the customer. He doesn't like it? Wants a bunch of tweaks? No big deal, we didn't spend much time on it and can reiterate as necessary. A better experience all around.

As a software developer myself, I see nothing wrong with that. I've spent a lot of years tediously grinding out code that does some essential but pretty boring stuff. Now I just get the AI to do the grinding. These days it does an amazingly good job with little effort on my part, and I'm getting better results with fewer prompts than just a few months ago. The improvements over the past year have been incredible.

I do understand people's concerns about the skill pipeline though. I know what components I want the AI to build, how they should hook together, how they should be tested, etc. That's mainly because I used to have to do it all by hand. But I think as time passes even the architectural details of many applications will become boilerplate that the AI can easily handle. Project managers will define requirements, the code will be generated quickly for review, there can be multiple iterations over a few days if needed. The time from idea to product will be vastly compressed.

As the Times article says, “The blessing and the curse is that now everyone inside your company becomes a coder”. That ain't such a bad thing in my opinion.

>> meanwhile it lost track of previous requirements more than once and wiped that out

It's best to start off by telling the AI to write an implementation proposal for your project and get it to put all your little requirement details in that. Then you can tell it to implement the plan in phases. Revise the plan later if necessary. That way everything is documented and the AI knows exactly what to do.

Why did they make him CEO if they know so clearly that he can't be trusted?

>> What if the white hat AI introduces the vulnerabilities?

Always possible of course but I find that the LLM's are better at writing robust code than most humans. Yesterday I was working to make a basic login page for a web app. After I got it working I asked the AI how I could make it more resistant against hacking and it came up with a long list of improvements. Brute force protection, cookie security, session binding, idle timeout, concurrent session limits, login anomaly detection, etc., etc. Very helpful.

It's a hassle to pick and choose from the various models I agree. I'm using the Windsurf IDE and recently they came up with an "adaptive" model picker that supposedly sends your prompt to the best model for the task. I was using it yesterday and got good results, no telling which model or mix of them were doing the work.

I am thinking it costs them a gigantic amount of compute resources to run Opus 4.6. In my experience though, it is the premium model for coding and in many cases worth the extra price.

Claude can be slow I agree. But I submit that is because so many people are demanding to use it.

I work with other models too and get good results at times but if it is something particularly difficult the Anthropic models tend to do a better job for me.

"In the last few weeks, Mythos Preview has identified thousands of zero-day vulnerabilities with many classified as critical."

We are moving into a scenario where there's a race for extremely capable white hat AI to identify the existing vulnerabilities and try to plug them, and black hat to find and exploit them. I think this is a good move to try and get the white team ahead of the game. There's a possible apocalypse here.

I generally use Sonnet, it is very capable for most things and cheaper. If it gets stuck or starts to struggle I switch to the better model.

The Claude models are the best by far for coding assistance in my experience. Apparently a lot of other people think so too because Anthropic is getting swamped. They are having to ration out their compute resources and in some cases have raised their fees to 2-3 times more than the lesser competition charges. I'm finding that in order to keep costs down I'm having to use 2nd-tier models for simpler work and revert to Claude for the heavy lifting. A hassle.

Clearly the demand is there. At this point I expect Anthropic is revenue-limited by their infrastructure availability so it makes sense that they recruit the big players to help beef it up.

>> Was giving you an out

That's an obvious lie, and I stand by my statement.

>> you showed nothing

Another obvious lie. I merely showed something you didn't like.