P03 Locke
- 0 Posts
- 49 Comments
21·3 days agoThe only systems to have level 4 are taxis that only operate in limited areas, on limited, known routes, and they are well known for being very easily confused and blocking emergency vehicles.
Even then, they have overrides for human operators when they fuck up. Driverless cars, my ass.
That’s okay. We’ll take a stand to vote in a bunch of Democrats, have a majority in the House and Senate, get blocked on presidential veto powers and filibusters, and the public will just blame Democrats for not changing things. Because they are too fucking stupid to understand basic federal government processes that you learn in elementary school.
And then two years later, they’ll betray their own country again and vote GOP.
Even when voting doesn’t take much effort, the apathetic dipshit public can’t even be bothered to stay motivated for more than one election cycle. Meanwhile, a bunch of old ass boomer conservatives are clocking in at the polls every fucking year.
They are too scared to name their own bosses.
If Starlink wasn’t run by a fascist supremacist corporate dickhead, they’d offer their plan for fucking free, to show their support towards Ukraine. That was the whole fucking point. This was going on right when the US and Europe were offering military and weapons aid.
Instead, Elon Musk took one look at Rule of Acquisition #34 and ran with it. He used it as a PR stunt, and took it away to ask for more money.
You called them “murder drones”, like you disapproved of Ukraine using drone warfare in their fight to defend themselves. That’s tankie behavior.
7·4 days agoOf course it’s an unpopular opinion among the vehemently anti-AI crowd over here, but you can get a pretty good headstart with LLMs, local ones even.
The thing is that you don’t just take what is given and immediately send it back as a shitty PR. Code review it, figure out what works and what doesn’t, have a test plan, have unit tests, stand up a sandbox and look at the frontend results, use a second agent for a blind code review, etc.
Hell, if you aren’t spending at least the first several hours asking questions and crafting a complete design plan, you’re not doing it right.
Tankie much?
I predict a lot of H100 GPUs are going to crop up super cheap on Ebay pretty soon.
Yeah, and with the state of cybersecurity, we can regularly experience the beginning of that movie Upgrade!
If I could pick a company that has fucked up more than Microsoft or EA, it would be IBM. For the past 20 years, the whole company has been on a wild bender of buying good products, and shoving them headfirst into the ground.
Nvidia is the shovel maker. They don’t care who buys the shovels.
Chinese who don’t say these opinions are labeled a dissonant and oppressed by their government. Just like how elections are completely staged in dictatorships.
This graph is completely useless.
Glm 5, which i think is near frontier model capabilities, takes around 1.5 Tb of vram un-quantized.
DeepSeek’s latest version is 508GB, and even that is wasteful, compared to the wild difference in sizes with some of the quantized versions. The huge amount of parameters and weight sizes are completely unoptimized in terms of sizes, and there’s all kinds of diminishing returns with the haphazardous way these LLM corps are applying new model versions.
Messy brownfield code requires a lot more meta-analysis from previous sessions, just to get all of the spaghetti code into a manageable set of rules and documentation.
Don’t just send the agent blindly into the code. Every session, the agent is just going in fresh without any context on how the code works. Break the codebase down to a 10,000 foot view as a summary, and then get more documentation and agent rules for each of its major sections. Good tests help a great deal, too, and if there aren’t tests, start building them. Start creating skills just for the codebase, or any common tasks you’re asked to do against it.
Also remember that you don’t have to write any of this stuff. Have it write the docs, rules, skills, tests, and just review it to make sure it looks right.
That seems like the worse of both worlds. About 95% of my code nowadays is agent-generated, but it’s based on my input, because I’m the fucking expert. Outsourcing the work just turns it into the same “garbage in, garbage out” process that outsourcing always does.
I really wanted to spin up my own home LLM server, but the prices around that are still scary right now.
This is assuming a good faith argument that Trump actually gives a shit about public money. And frankly, even saying “good faith” and “Trump” in the same sentence already makes me sick.
It’s not his money. Trump does not give a shit if the money is wasted. He only cares about propping up this AI bubble for as long as he can, to keep his own stocks afloat.