lbreakjai

You just wait, we're still early®.

If Claude is good enough to review the code, then it should be good enough to write it right the first time, isn't it? Why waste money on CI?

Not when the product you're selling is to make humans obsolete. Imagine if Apple had a budget to buy pixel phones for every employees.

Friction is good. I've been in this for fifteen years, and I can't name a company whose product would have been better if they somehow had a magic wand to instantly implement even half of those 5 years, 600 tickets deep backlogs.

> The AI usage still needs a knowledge worker to direct it in ways that it can't direct itself

So we're back to square one, and your productivity increase has a ceiling that's much lower than 5x, let alone 20x. There's a limit on how many things one can keep track of, how many code changes one can reason about.

I disagree. If a model is good enough for your task/workload, then subsequent smarter models won't really help you.

I'm not surprised. At work, we paired devs and non-devs together, and we're sharing subscriptions, since we tend to exclusively use one or the other. It's like when Copilot was billing per query rather than per token, always bound to disappear.

> and is probably difficult to update (original writers long gone)

The new program won't be any easier on that front. It literally doesn't have an original writer. So the new program will certainly be maintained and updated by a LLM, at which point I wonder if the language switch mattered at all?

The end user always pays, one way or another.

They've been rooting for ours, it's only fair.

The amount itself is probably meaningless. The fact that there was a shortfall at all is, for me, the signal.

Not too long ago, people were fighting to throw money at them. That might be the sign that we're past the peak of inflated expectations and going down towards the through of disillusionment .

This is such a terrible failure mode. I had quite a few of those "People are saying that ..." that turned out to be sourced from the only two reviews left on the product I was looking for, but made it seem like it had broad consensus.

Cost won't become prohibitive, it'll trend towards utility.

I personally have dumped everything into money market funds. That should give you somewhere between 3 and 4% return assuming you're using a dollar one.

Except the other models/harnesses can already search the web. The only reason why I still end up on google is if I need to check the spelling of a word ("Didn't you mean xyz instead?") or for very quick and small checks ("Number of inhabitants in X city")

If I need to _really_ search, something that would take comparing multiple links, I go to Claude/ChatGPT directly.

Just yesterday, I was looking for a small tool, I didn't really know the name, but had a picture. My first reflex has been to paste the picture in Claude, and ask it to find me places where I could buy it, that delivered in my country.

Yes, but google makes money from the ads all around the results, which I won't see if I'm using a Claude's UI.

Freedom never meant that the society much be changed to accommodate you. You're free from persecution (Conditions apply), you're free to behave the way you want, to associate with who you want. Likewise, Society is free to not associate with you and move in a different direction.

I don't think the analogy holds. AI isn't automating one job, it's automating _knowledge work_. There's no factory work to flee to, literally and figuratively.

Try to come up with a job that both doesn't produce anything physical, and can't be done by AI.

For a long time, this was something you couldn't really do on an iPhone. You couldn't just install Firefox with uBlock. Even ignoring the ads, most websites seem to purposefully make the experience miserable to push you to their app.

I had to double check when I saw the "piano keyboard" demo picture. I assumed I was on a parody website. How the mighty have fallen ...

This is an often underrated problem. I used to work at a company where we would hold demos every other friday, with drinks and nibbles. We would put a lot of thoughts into what we were going to present, and were really careful about the messaging, because execs and sales were watching.

The last thing we wanted was for an "Isn't it neat?!" hacked-together demo to be mistaken for something shippable, if workable at all.

It boggles my mind. I only use my phone as a very last resort. It's such a terrible experience overall. Between modals, banners, and ads, it's downright impossible to even read the content you want.

It's not even a screen size issue. I couldn't imagine diving into anything remotely serious without being able to open a new tab and search a specific term/concept/fact in a quarter of a second, without having to fight the autocorrect at every step of the process.

> Did Microsoft's chief branding officer move over to OpenAI?

Obviously not, otherwise the four models would be called Copilot, as well as the harness.

Sure, but the message I was responding to was describing casual uses: people using AI to argue with their doctor or review some contract, or what have you.

The problem with the "double the work" narrative is it ignores the second order effect. You can do twice the work, but so does everyone else. You doubled the supply, was there a massive pent-up demand for lawyers or accountants at the current price point?

Indeed. On the other hand, I fixed my underfloor heating with the free claude version, so unless they remove the free tier, I still have no reason to pay for it.

It's quite a leap to assume any kind of issue would require a $100 subscription. Plumbing, mechanics, electricity ... The issues are usually far easier to trace and debug than in software, because the systems have been designed by smarter people.

You're not buying some arcane and complex solution, you're buying the convenience of not watching a couple of youtube videos or articles of someone troubleshooting the exact issue you're having.

That's true, I do all of that, but I'm using the free Claude version. It's been more than enough to handle the few home renovation projects I'm currently working on.

That's quite a leap to assume there's a ton of money to be made with the casual users sending perhaps a couple of messages a day to real an email and draft a response.

The speed of change has been astoundingly fast. We went from temperatures above 30 degrees being newsworthy, to months on end with temperatures above 30 degrees in, what? Four years?

Everyone I know is in the process of getting A.C installed but installers can't cope with the level of demand. There's some ideological resistance, but understand the change happened as close to overnight as you'll get when talking about the climate.

I still very much love DataGrip, but it's the last of their IDEs I'm using, after being a Webstorm and IntelliJ enthusiast for ten years.

That's a very "Spherical cow in a vacuum" approach. You would need to accurately model the boss, as well as all your interactions with the boss up until your raise request, as well as the boss' environment when you make your request.

You could be on the wrong side of a reasonable boss, or in his good graces but during a bad day ...

That's the point the writer was making when he mentioned role playing games. In isolation, you could spend hours drafting the perfect plan, that wouldn't survive two minutes of contact with reality. You just don't know if your model is correct until you try.

The nomenclature is quite loose. Speculaas and speculoos will be used interchangeably in parts of Belgium. "Spéculoos" will refer to both in the french-speaking part, while I anecdotally noticed Flemish people will use "speculaas" even for biscoffs.

That's why I only eat kruidnoten, less of a linguistic quagmire.