It would also be incorrect. "Up to" doesn't mean any possible upper bound, it describes the top of a known range. I can't say a car I'm selling goes "up to" 900mph if it tops out at 120mph.
¹ drivers are responsible to get the car into a low earth orbit themselves and need to be familiar with the orbital maneuvers needed to accelerate the car to 900mph
Funnily enough, for me AI writing is usually the summarizer rather than expander. I would usually give Claude 2-3x the amount of data in notes, context, braindump etc; when doing person-to-person communication, using AI to _expand_ the context rather than shrink it is doing everyone a disservice.
At some point, the rabid anti-generative AI crowd becomes tiring. This is extremely masturbatory and very intellectually lazy piece of writing whose only purpose is to raise cheers from the same echo chamber an author is in.
It's a reaction. Nobody would be "anti-AI" if AI wasn't being shoved into everything while at the same time shaming you like you're stupid for not enjoying literal slop.
Sometimes I get code handed to review that has to same variables set multiple times, additional classes/methods/functions added that aren't used, and when questioned I get a shrug as a response.
Slop.
Sometimes they review it and tidy before handing it over. Much like although we have printing presses, hopefully someone reviews it to make sure that they actually printed square on the page.
Not slop.
The problem is that we have lazy people and mandatory LLM usage has forced us to confront these people that don't even proof what they write. Before they would just hide in the background and not do anything.
No, this isn't what I mean. What does the word "literal" mean to you?
Because to me it means the opposite of figurative.
Ex., if I say "It literally costs an arm and a leg", that means I'm really paying in body parts.
Or if I say "Your book bag literally weighs a ton", it means I'm saying it's actually 2000 lbs. and would take heavy machinery to lift it.
Or in this case, you are saying the "slop" program or code somehow actually transformed into some mystery food melted into a gelatinous dish when you add "literal" to it.
Given how bad some of the models do on somewhat similar problems, I'm sure pelican is included in training set now.
Similar problems - given airplane outline and implementation constraints do painting scheme (constraints something like "it will be implemented using covering film, hence no gradients, no impossible cuts, not more than 2 colors on engine cowl, etc). Google Gemini is meh, but GPT models are just terrible, don't have Anthropic subscription at home, hence have not tested.
Bad pelicans are in the training set because it's read his blog post. Including a good pelican in midtraining wouldn't help the problem because you'd just produce that every time.
Based on the name, he's either Ukrainian running away from the war, or Russian, running away from being complicit in the war. Shitty situation to be in either way.
Necessary context for someone probably being more detached from US politics:
Former FBI Director James Comey — fired by Trump in 2017 and one of his most prominent critics ever since — was indicted this week by Trump's Justice Department for allegedly threatening the president's life through a May 2025 Instagram photo of seashells arranged to spell "8647" (a protest slogan where "86" is slang for "get rid of" and "47" refers to Trump as the 47th president), which Comey says he intended as political commentary, not a call to violence.
A snide and vitriolic remark that observes on how the first paragraph actually addresses the concern of the person which hasn't read the article. A further continuation on this being representative of the state of modern online discourse.
In consumer space, 69 KB of RAM (138 x 4 kbit chips) would cost around $1700 70s dollars for the entire package, ~$10k in modern dollars.
Radioationed hardened for space though — $50k-$100k in 70s dollars, roughly the price of a Silicon Valley house back then - $300k-$600k in today's money.
Lead makes things worse, not better. High-energy particles go straight through a couple mm of lead no problem, and lead itself is radioactive anyways. The problem is when a particle punches straight through a chip, leaving some energetic charge behind. You won't stop that with a paper-thin layer of lead.
Um, I guess in that there are naturally occurring isotopes of lead that are unstable. But those are very rare and can be removed. By this standard, the chips themselves are radioactive since they are made of silicon. By this standard, you, everything you eat, every plant and animal is radioactive since there are trace isotopes of carbon that are radioactive (that's how carbon dating works). And sunlight is very radioactive by this standard.
However, the materials we actually use in chips is highly processed and radioactive isotopes will likely be removed if a centrifuge was used in any step in concentration process. Likewise the lead used in a space shield will probably have similar processing of the material
"Also, lead is extremely dense."
This is the real reason lead isn't used. There are plenty of other elements that shield ionizing radiation quite well and are less dense. However, they are also more expensive than lead which is why we use lead on earth for shielding (its cheap) and use something else in space (where density is more important than price).
Because lead turns a single high-energy particle (that would disturb a single bit and punch through) into a shower of many low-energy particles (that would disturb many bits AND induce lattice damage over a wide area).
