Definitely not. You need 4 digits for that. :)

But that's the thing: All of that applies equally to human generated code, too. I've seen plenty of humans do all of those things even without AI assistance.

Where AI today consistently beats a human is on speed. No human can generate code as fast as an AI can. But speed of generation isn't everything. Right now (competent) humans have the edge when it comes to seeing the big picture, anticipating the code evolution over time as requirements change, and stuff like that. But it's hubris to think that human code is always better than AI code. Yes, I've generated a lot of AI code now that wasn't that great, where the AI code wasn't handling some condition that it needed to or wasn't anticipating some future change, but if I'm being honest, I've also frequently experienced the opposite, where I asked it to generate code and it handled some condition that I would have forgotten about if I had written it manually. The AI isn't perfect, but people aren't perfect, either.

AI code generation is here to stay. A competent human, working with an AI, can easily outperform a human operating alone, much like two humans using techniques like pair programming can outperform a single human when it comes to producing reliable code. Working with Claude Code feels a lot like working in a pair with a partner that is ridiculously fast.

If open source projects reject this outfight, it seems foolish to me.

DST is the summer time, so the plan is to keep summer unchanged and change winter.

Software development is changing. I started coding in BASIC in the 1980's and have been coding now for over 40 years, over 30 years professionally. I'm good at what I do, but the AI is faster. Claude can churn out code faster than I can, and it's often better, catching some conditions I would have missed. That said, it often messes up, misses the mark, or goes in directions that aren't right for the larger context in which the code exists.

Today, professional software development is best done by AI with skilled human guidance and review.

Rejecting AI generated code in today's environment is trying to turn back time. On the other hand, rejecting a submission where there is no human who can "understand, own, and be able to fix the code they submit" makes perfect sense. There is a big difference between asking an AI to generate a fix and blindly submitting the first thing it spits out, versus having an extended session with an AI, correcting it where it goes wrong, vetting and testing the patch with human review and testing, then submitting the PR.

I haven't been in one of the new ones, but I own two Bolts (2020 and 2021) and my dad has a 2023 Bolt EUV. They don't have "dumb buttons for hot, cold, and fan", they have a thermostat where you set the desired temperature and then it controls the heat based on that. The computer makes some counter intuitive choices at times, turning on the AC when it's already cold or the heat when it's hot. You can override it, but it's not the direct dumb buttons and knobs my older cars had where you directly controlled the air flow with mechanical levers.

Still great cars for not too much money.

It really doesn't, though. Even without FTL, given enough time any technological species should eventually spread to all space.

Cool Worlds on You Tube covered this recently:

https://www.youtube.com/watch?...

With sublight speeds it takes millions of years to spread, but we've had those millions.

In Altered Carbon they grow human bodies without a consciousness and then rather than harvesting the organs for another person, they move the other person's consciousness into the body.

I've been using GitHub Copilot for a few months now and I can say that it definitely saves a little bit of time. I've seen reports that it makes developers 33% faster and it really does not come anywhere close to that. Where it shines is being able to fill in some boilerplate code faster than I can type it. When what I'm writing is something where the next few lines are something that an undergraduate student could easily predict, it can fill it in and I can tab complete it faster than I can type it. (Even when it's wrong, it's often faster to accept the close-but wrong answer and fix it than to type it all out manually.)

Expecting it to write whole routines or do any sort of serious architecting is going to be a let down.

But here's the thing: Copilot for Business is $19/month. Compared to a typical developer salary, how much time does it need to save per month to justify that cost? If it saves minutes per month, you're probably still coming out ahead. If anyone really believed it made developers 33% faster, they'd be charging a lot more.

But it definitely requires some skill to use it well. It comes up with some truly monumentally bad suggestions sometimes, and often suggests things that superficially appear correct but aren't. You need to be able to know the difference. If you don't, I could easily see it making things worse.

I was thinking of this one. Or even this one.

It's easy to forget just how much higher the moon is compared to the ISS.

There's a graphic on Wikipedia that shows the radius of the earth (blue ball), the ISS orbit (purple circle), and how far out the moon is.

Saying that you can reach the purple line doesn't really mean much in terms of reaching the moon. It's literally 0.1% of the way there. (About 400km to the ISS, vs. 400,000km to the moon, give or take.)

For a long time, Debian has been striving for Reproducible Builds, where building a source package produces the same output every time. Even with a local NPM registry, it would be nearly impossible to keep track of what was in the registry at the time of doing the build unless that content is actually in the source package itself.

I voted by mail in New Jersey. There was a ballot that said who I wanted to vote for with no personal identification that went into an envelope with my name, address, and signature. Someone had to compare my illegible scrawl of a signature against an illegible scrawl on file and use that to determine if the ballot was actually mailed by me. With PKI in place, we can replace the hand signature on the envelope with a digital signature, which is much more easily verified consistently. Hand signature verification is a joke. There's no way to catch forgeries without throwing out lots of legitimate ballots. PKI can fix that, at least.

Most languages have a no-op statement. C allows just a plain ";" as a do nothing statement. Even going back to Fortran, there was CONTINUE. It's a very old concept.

I had been pledging $1/month to several different creators. With the new fee structure, it's better to only fund one creator each month and rotate that creator every month. That's ridiculous.

I canceled all my pledges this morning in protest.

That's actually not true. When you look at the Lockheed Martin Compact Fusion Reactor, it's being designed to be small enough to fit on an airplane. It's a lot bigger than a "Mr. Fusion", but compared to a typical fission reactor, it's tiny.

"Hacker's Keyboard" has a number row, tab, and arrows.