HN user
Nice! I'm surprised how snappy this browser editor is, especially once I start doing really tiny, like 0.3mm large maze fills and other tricky stuff. I do a lot of generative art (not AI) stuff, and have a plotter lib for rust based plotter art (aoer-plotty-rs), and I am going to steal a few of your features (maze fill and fermat spiral fill).
Is it that they are copying existing stuff, or that success is biased to the attempts at something that has a lot of existing examples to plagiarise?
I would propose that plenty of people are trying to use LLMs to build unique new projects, but without having been exposed to solutions to those problems, the LLMs are far less likely to succeed.
I used svg2polylines to load the SVG into line segments, interpolating any splines into something easier to use for a plotter. Feel free to take a look at the core/project/import.rs in the project I linked. You might also want to look at the core/post.rs which does the gcode generation.
I am inordinately amused by the fact that Microsoft's Github has suspended access by Microsoft's Azure (and anyone else) to their Microsoft code-base because of a TOS violation.
Really drives home this org chart: https://www.businessinsider.com/big-tech-org-charts-2011-6
I wrote a pen-plotter GUI and gcode sender in Rust. By hand. Like an animal.
I am the only user. Sometimes it's the process that matters, and exercising your brain is important too. I get that there is a lot of existential dread around AI taking our jerbs, and excluding humans from the process of creative work, but... you can still just write code, just for the personal satisfaction.
/me looks at the `build.rs` file in my Rust's project's `Cargo.toml` and laughs nervously...
(For non Rustaceans: "Placing a file named build.rs in the root of a package will cause Cargo to compile that script and execute it just before building the package.")
It would absolutely have been valid to ask that question of the inventor of the tractor too.
It's even more relevant to ask of the CEO/CTO/COO/etc. of the companies that are selling hard on eliminating humans from as many workflows as possible.
If a headline asks a question, the answer is always "NO".
Ahhh crap.
While I am sure there are stylistic reasons for using that color, there is another common reason why you see blue-green colors in paint, especially in older industrial environments: zinc chromate/phosphate corrosion protective coatings. Zinc chromate primer is the color you see on the interior surfaces of some aircraft, to inhibit corrosion. Zinc phosphate is more of a gray in most cases, although varying paint chemistries result in a spectrum between those two, with seafoam nearly smack in the middle.
These are still available today, although the chromate version seems less popular for general use due to toxicity, especially (I assume) in the case of a fire.
I have painted quite a few bits of sheet metal with a sea-foam-ish blue-green/gray paint back in the day (30 years or so ago). I don't recall the manufacturer, but it was a zinc conversion coating in nearly exactly that seafoam color, which has probably stolen at least a few years of my life expectancy. The same company sold other paints in a sickly mustard yellow, and close to fire-engine red, all with slightly different chemistries, I assume for different base metals.
To clarify my thinking on this, I don't expect that we are going to end up in an air OR land war with the USA anytime soon (and honestly hope we won't). I do expect that Trump, for as long as he occupies office, is going to exploit anything and everything that he perceives as a source of leverage. If he can manufacture a pretext to force a renegotiation on maintenance, parts, software updates, and any/all other operational costs once we sign a delivery contract, then that's what he'll do.
The F35 also leaves Canada vulnerable to US policy, in as much as we're risking autonomy around any decision we make that isn't aligned with US interests, as long as we're dependent on their planes.
Canada shouldn't buy the F35. The Saab is a less capable plane, for sure, but it doesn't leave Canada (and it's defense) dependent on the US at a time when the president is openly floating the idea of "acquiring" Canada.
I am somewhat confused by this post. If the AI assistant is doing such a bad job that it lights up the linting tool, and further, is incapable of processing the lint output to fix the issues, then... maybe the AI tool is the problem?
If I hired a junior dev and had to give them explicit instructions to not break the CI/lint, and they found NEW ways to break the CI/lint again that were outside of my examples, I'd hopefully be able to just let them go before their probation period expired.
Has the probation period for AI already expired? Are we stuck with it? Am I allowed to just write code anymore?
This paper runs very parallel to Dunning-Kruger, and it's surprising that nobody here has commented on it (although the paper itself DOES reference Dunning-Kruger). It's a bit sad, really. When you realize that most people that buy into this stuff are trapped by a lack of cognitive ability, rather than being rooted in malice. There's a dose of https://harmful.cat-v.org/people/basic-laws-of-human-stupidi... in there too, as even well educated, otherwise intelligent people can get trapped in these conspiracy theories.
"Conspiracy believers not only consistently overestimated their performance on numeracy and perception tests"...
Compared to:
"The Dunning–Kruger effect is defined as the tendency of people with low ability in a specific area to give overly positive assessments of this ability."
That 68k/yr wage only sounds good if you're still thinking in circa year 2000 dollars. Nobody is making the mortgage on a house on 68k/year, and they're not starting a happy family if they have to do 20+hrs/week overtime in order to turn 25/hr into 68k/year. I remember earning nearly exactly that wage back in the early 2000s, and barely making ends meet in a cheap rental, so it's certainly not a great wage today.
BBC is reporting that the Army has confirmed that it was a UH-60, not a VH-60, ie: not a VIP transport: https://www.bbc.com/news/live/cy7kxx74yxlt?post=asset%3A62b9...
Although it's early on and these communications are often chaotic/inaccurate.
Executive whose dominance is dependent on a moat (cost of training large models), points out the importance of that moat...
Of _course_ he wants to discourage competitors, but that is PR, not technical commentary.
The problem with LLMs as a replacement for StackOverflow (and any other peer reviewed Q/A site) is that the LLM has, shocker, no peer review. Combined with the fact that the user has insufficient expertise to adequately vet the responses (ie: they needed to ask in the first place), there is a trap where the user can end up using an incorrect response without any expert feedback to help them make an informed choice. As the LLMs get better, I expect that this will improve, but for now, it's gonna train a lot of novices to apply bad practices.
As a comparison, here's the "Why RON" for the native RON/Rust format:
Note the following advantages of RON over JSON:
* trailing commas allowed
* single- and multi-line comments
* field names aren't quoted, so it's less verbose
* optional struct names improve readability
* enums are supported (and less verbose than their JSON representation)
This study doesn't propose anything that can actually move the needle for food security. It requires animal based feedstock for the Pythons, which kills the efficiency claims because you need to grow an animal of mass X in order to obtain another animal of LESS than mass X. This is just a slightly more efficient method of recycling leftover biomass. It'll never be viable as a primary protein source.
If they had suggested a viable cold blooded herbivore, we might get somewhere.
New car reliability measured in warranty returns per vehicle is much WORSE over the last few years than 15 years ago. Turning your nose up at the 15 year old Toyota is probably making a choice to be _less_safe, since at least the old Toyota's recalls are known, and likely fixed. The new vehicle doesn't have enough time on the roads for all the dangerous defects to be identified and recalled yet.
If you're on a 737-max-anything, then worry a _tiny_ bit, otherwise, no. All their other planes have excellent safety records.
For comparison, Russia had a 2.29 fatal accidents/million departures safety rating in 2021, which is terrible, worst of all operating regions. 737-maxes had a 3.08 in 2019 when they got grounded. Every other plane in North America is averaging about 0.20 right now.
Granted, the 737-max's stat is based on a fairly low number of crashes. Regardless, the odds of a fatal crash are still vanishingly low on any given trip.
A switch on a wall with an arduino in it where you flip it and DNS is updated to point there & a message is displayed to the users.
If you want a Klugey non-production IoT solution that bodges up something really important THESE days, all the cool kids are using ESP32s.
And as much as I think that would be a totally inappropriate solution for src.ht, I kinda wanna go make a "black-hole" switch for my office.
We upgrade our distros pretty much fearlessly, all the time. While I have had breakage from Kernel upgrades, they've been very rare (and generally related to third party closed drivers). Kubernetes is _not_ more complicated than the Linux kernel, but it is much more dangerous to upgrade in place.
Not sure I trust these guys to do long term support, even after they promise it: https://killedbygoogle.com/
Hell, I don't even trust that they'll still own/maintain _Android_ in 7 years.
I never mentioned reinforcement learning, and my DK statement was completely around using flawed fonts for graphic design, etc.
My partner _is_ a professional graphic designer, and we _have_ seen some pretty terrible client graphics that came out of Midjourney. They're amazing for what they are, but it's very difficult to get something out of it that competes with a professional illustrator, even ignoring the whole copyrighted content in the model issue.
OK, that's cool, but those fonts are all terrible. The serifs are all different sizes and shapes, sometimes on the same letter. The kerning looks like a random walk. The stroke widths are all over the place, and/or the hinting is busted.
Now, that said, it's pretty amazing that this works at all, but it'll take some pretty specific training on a model to get something that can compete with a human made font that's curated for good usability _and_ aesthetics.
Sadly, we'll also probably see adoption of these kinds of fonts (along with graphic design, illustration, songwriting, screenwriting, etc)... because "meh, good enough" combined with some Dunning-Kruger.
TL;DR: Thanks, I hate it.
It's a neat thought experiment, but the underlying assumption:
The world's energy consumption was about 5.67x1020 Joules in 2013.[18] This number has increased by more than 2% per year on average in the last 50 years. The average world economic growth rate in the last 50 years is about 3%, which requires a corresponding increase in the energy supply. So, the 2% growth rate for world energy consumption should be a conservative assumption.
Four billion times as much energy as we do today. I don't think there's much risk of us growing our population to that degree, nor of us being that power hungry if our population stabilizes. We'll be either extinct or back to a sustainable agrarian population far before we reach that upper limit. Honestly, if we produced that much power, I suspect we'd have long since boiled the oceans, making the whole argument moot.
TL;DR: Don't extrapolate FAR into the future based on a small (relatively) set of data points.
This is hardly a new problem though. Thieves used to be able to pull the wires from under the steering column and start a car.
For the more serious attacks mentioned, the automakers are essentially saying that they build extremely vulnerable systems and are afraid to disclose that fact.
I suspect we're in violent agreement though, that the correct outcome is to fix the vulnerabilities AND to document the repair methods.
That doesn't return your money to you like the credit card though. The irreversibility 'feature' of most cryptocurrencies makes it really hard to recover funds stolen through fraud... I guess that is a desirable feature for anybody performing fraud though.
They could just have operators translate the configs automatically on the fly
How do they test that across the infinite number of permutations of configurations and deployments of K8S in the field though? It'll work for some people on the happy path, but it's really hard to maintain over time. Worse, it'll break randomly at some point in the future that is hard to predict, instead of at some publicly announced point in time where the breaking change is deployed (how it happened this time).