Agents are nothing more than magnifiers for the operator. They 10x the user.
If you find your agent sucks at programming and has zero taste for good design, then that's arguably a 10x of the operator.
I'm quite pleased with the quality of programming and design that I'm able to achieve with the processes I do with agents.
I've certainly noticed that when I watch a junior use AI they've produced nothing but the most intolerable slop man has ever seen. And I've seen principles and architects produce one shots that are such shining examples of quality engineering that they remain nothing more than goals to me.
Throughout my career I've met many engineers who drew a line in the sand and said I'll never work at a business who does a practice that accelerates the business and lowers the quality of the engineering.
Most of them exited this industry before their career was over.
It's arguably the most successful piece of software in the past decade.
It took a software company from a 10 billion company to a 1.5 trillion company in about one year.
This is the counter example to you. With AI, a small team can make a trillion dollar business in the order of days.
Hard to care about developers crying about the usability of their Dev tools when their businesses are 100Xing and 1000Xing per year on the back of AI code
And you have no idea if the Microsoft outage was caused by AI. You are just lazily attributing an issue to AI because it's convenient for your argument which is a very ignorant tactic
I don't mind the app functionality of my air purifier. I get a push notification when air quality degrades, and I can see data and charts of air quality over time. I can change the settings remotely, which is nice because I have a small child and keep the "lock" function set for the actual controls.
It's easier to setup and maintain schedules, routines, or more complex functionality.
It's certainly not critical functionality, but it's nice. Same with my laundry machine and dryer sending a push notification when the cycle ends. Unnecessary, but nice.
> It's easier to setup and maintain schedules, routines, or more complex functionality.
Complex like wanting to hit the "Good night" button in Apple Home and have it turn the purifier to full blast for white noise, the workflow in TFA that wasn't possible with the app experience?
It's one of those things quite interesting to see the difference in response from one consumer to another. I can't stand the washer/dryer notifications and finally turned them off a few weeks ago. Especially when I'm on the road half the time, why should I care my wife did 8 loads of laundry in a day? When I'm home, the chime is enough to hear from the other room anyway and the notification on the phone usually just backlogs and gives me a load of red badges I need to clear at the end and beginning of every day for notifications I didn't even get because the phone wasn't with me and I changed the laundry anyway.
For an air purifier, I feel like I'd probably want to see outcome-level statistics convincing me it actually matters to lifespan, healthspan, quality of life, something, anything at all, to know and adjust minute-by-minute the exact air quality going through a purifier, or for that matter what the purifier is able to detect even reflects the quality of air elsewhere in a house with a whole lot of air volume not going through that purifier very regularly.
Totally disagree. I do not think local models will displace cloud models any time soon, and I think distillation has been demonstrated to NOT have better performance and is LESS efficient. Chinese models use dramatically more tokens and time to achieve dramatically inferior results. Watching deepseek spit out 20M tokens when Opus solves in <1M is always an entertaining window into "efficiency" and "performance".
Local models literally cost more in electricity than using far more powerful cloud models cost in inference.
When your options are "dramatically drain your iPhone battery for substandard AI, or have instantaneous cloud AI with no battery drain" what do you think the average user will choose?
And that's before we discuss how ridiculous the pricing is for home hardware that has a chance of running a halfway decent model at a halfway decent speed. Residential electricity + overpriced underutilized hardware means you're going to pay 100X or more per token.
I think the future prospects for local AI are looking worse than ever. It will be far more expensive for a far worse experience for the foreseeable future. It will fill a privacy/control niche, but beyond that, it's not a serious option
Code review is a problem only because it is a backstop for everything else. We skip architecture review, we skip prototyping, we one-shot a result in isolation from our team and then expect a human at the end of the line to do our job for us??
Code review is great if the code isn't a surprise, the architecture was already agreed on, the AI reviewers are already happy, and you just need a final pass.
Code review was already terrible before AI if you were slopping down weeks of work on a poor teammate to sift through.
I tell the younger engineers at my company the same thing today I said 5 and 10 years ago. Make the review easy for another engineer to do it, and they'll do it. Make it hard on them, and you won't ship shit.
But the "make review easy" part is one where I'm failing (a bit) to keep up, I feel.
Also, what does "a final pass" look like to you? E.g. if a PR is 3000 lines... I'd guess one approach is enforce smaller PRs. I've opened a few, looked at the line count, and closed them, to be redone/broken into stacked PRs.
If I'm reviewing a PR that I'm familiar with the plan/work on, my "a final pass" is often just "read and suggest touch-ups of the docs". But that's really just the first step of a "full review" too, so it skips other stuff, and I know that isn't a proper review in the back of my head, but on to the next thing! :|
(Because of several factors including joining a much less mature org git-wise this year, I'm full of internal questions/dilemmas around PRs that I'm working on feasible solves for)
Code review complexity is always proportional to the associated work unit's complexity, not to some imaginary "keep PRs nice and small and nifty" complexity. I agree that consequences of bad SDLC ofter float up during code review, all the way up to formulating requirements and managing human resources.
For me, qwen 3.8 27b on my m5 max is so slow and dumb that I can't think of a single use for it. Haiku, Luna, or data center scale open models are ten times faster and ten times smarter (and inference so cheap that it's essentially free). It was a nifty toy to brag to friends about but why would I waste my time goofing with something so slow and dumb?
Are you insinuating that your agents are more effective at 20 tok/s than 200 tok/s?
I don't even understand your analogy. I'm not watching the youtube video or reading the book. The model is. And a big computer can watch 20 youtube videos faster than a small computer can watch 1, and do a far better job understanding all of them (bigger model at data center scale is both smarter and faster than smaller model at local scale).
Considering that the biggest and most successful play for SpaceXAI in the AI space has been "Sell compute to Anthropic" I also wouldn't say that all of the grok junk has been obviously successful, either.
"All top quality" did your agent tell you that?
reply