I think there might be some bias in the base model, but come on... owls are not a good thing to prove bias on. On a serious note, this was a discussed possibility pre AI with neural net weightings, so it would possibly make sense that the Teacher's weights expose their owl bias through sharing anything.

Claude has a very tenuous grasp of the relative path of any of its terminals. I check in and push to branch about every request because it can and has written meta programming scripts that seem almost reasonable that blow away the entire working directory. That is on top of the times that it literally just 'cleans up' the wrong path in one liners that you have to catch. Between Claude's refactoring metaprogramming vs. using given IDE(cursor) tools and that it often works around IDE protections in the terminal, it is kind of nuts to trust it for more than the one prompt. cursor also seems to promote some behaviors that are very dangerous as attempts to limit it parent directories or gitignore/cursorignore files encourages Claude to start grepping and listing files it can't get from the IDE.

Right now Claude and cursor are pretty dangerous for large projects, especially if it falls out of the context window and continues working after starting the 'probing' behavior for whatever legitimate debugging or testing it was attempting. It is incredibly dangerous if doing devops or file utility project work. Mocking everything in docker is also dangerous as it prefers to verify/run docker commands itself and then quickly forgets about docker and does those commands on the local system.

I would guess that a majority of competing Linux distributions were launched in the past 30 years from Redhat's complete lack of giving a shit about how, when, with what, and on what actual system their packages were made.

It seems that the copilot they sell for office/windows is super nerfed and barely better than asking bing. I don't know why they have taken so much effort to make it useless other than it must start spouting political opinions or complaining about their own software otherwise. Github was OKish, but now its kinda crap compared to the cursor.sh tuned versions of AI. Even cursor is a little uneven, but at least it gives decent code on well documented systems/APIs. It is about 90% hallucination free when you ask it about any Windows or Linux interfaces. Claude still seems the best for code and ChatGPT for planning/designing. Claude will still more time rewriting and ripping out functional code randomly(even when you ask it not to in rules and on the prompt), so you have to watch what it does but it is still saves a ton of research and development time.

Keep your source files to a single task/class and no more than 2-300 lines. Use meta-programming to manage code.

This has been the problem since streaming services started. Streaming services are like gyms, they want customers to pay but not to actually show up. On demand streams are pricey and actual regular users cost real money. Spending millions on advertising campaigns and new content has no honest way to calculate how many customers are added or retained with the investment. The whole business is luck and hope. Eventually there will be 2-3 streaming companies and they will share/trade almost all the same content have almost 0 growth and barely afford new content.

DJI will just make a Taiwanese subsidiary and/or a Canadian one just like all the 'banned' camera makers. They will have different cloud domains and charge 10% for the new custom logos on the boxes and shells.

This isn't that pro customer as it sounds. Appealing an arbitration to court is practically free in WA, so if you don't get the result you want you just wasted Valve's time and money on arbitration.

Yeah, we need medicine ran like the pentagon.

Or CEOs can be replaced by random decision making.

Well most modern buildings are built with your choice of cross linked polyethylene or polyvinyl chloride water lines because metals are too expensive today.

Most of these patents are expensive to prosecute and require reasonably smart attorneys to clean up and make more generalized and enforceable. The sad truth is that one in thousands is really worth anything ever and even then it is usually trumped up garbage like billions in tech. Most licenses in sane and normal industries are around 7% revenue for critical knowledge or processes. In more mechanical/industrial/chemical businesses patents are mostly design and are regular well rewarded and accounted. Heck, German patents require annual payment based on direct sale accounting of revenue versus some made up tiny token of cash and continue past employment. Tech utility patents are super broken and I think getting rid of rewards and therefore patents (you as an engineer gain nothing but liabilities and obligation by filing them) in tech is probably a good idea in general.

make cloud racks for Microsoft about a decade ago. They just couldn't figure out why someone would want a turn-key private cloud rack of blade servers with automated VM deployment of all services... If the same execs are there they can pat themselves on the back for only taking that long to learn how to sell cloud systems...

How many millions are unmaintained? Funny, I bet it won't be easy to find all those products/projects much less check for updates as easy as querying github.

They already strip quotes, remove conjugation/tense, and just about any other way to actually find a specific string of words in their search. I can only imagine what non-QA'ed barely related garbage their image search results will bring back.

Reminds me of supercomputer budgets in the 2000s. What did they run? Besides insane electric bills, just dick measuring benchmarks mostly. It would be interesting to know how they came to require that size of system and if they have done anything remotely big.