roxolotl

8 ms·

121.
by roxolotl 7mo ago
We don’t need agi or superintelligence for these things to be dangerous. We just need to be willing to hand over our decision making to a machine. And of course a human can make a wrong call too. In this scenario that’s what is happening. A
122.
by roxolotl 7mo ago
Yea so I’ve had an issue getting video output after boot on a new AMD R9700 Pro. None of the, albeit free, models from OpenAI/Google/Anthropic have really been helpful. I found the pro drivers myself. They never mentioned them. Th
123.
by roxolotl 7mo ago
It’s hard to align the two groups. As someone who used to prefer apolitical discussion I now find it very hollow to talk about <x> without including the societal implications of <x>. Like it’s possible to be interested in nuclea
124.
by roxolotl 7mo ago
The other nice part of huggingface’s setup is you can add theoretical hardware and search that way too.
125.
by roxolotl 7mo ago
Again that’s a nationalistic point of view. For someone unused to thinking about the world as “us” vs “them” where the designations of “us” and “them” are defined by national borders it can be surprising and seem like there’s missing inform
126.
by roxolotl 7mo ago
I think the point is “it’s fair to Americans that’s what counts” is a nationalistic statement. Maybe it’s the way to go. But it’s not refuting the parent who’s saying the missing piece is nationalism.
127.
by roxolotl 7mo ago
This is the data center version of https://xkcd.com/538/ . Realistically if there is a hot war what you’re saying seems accurate.
128.
by roxolotl 7mo ago
I don’t think this is what it’ll look like. Ads are going to be way insidious. One major power of these chatbots is persuasion. The end goal isn’t bombardment it’s going to be more subtle.
129.
by roxolotl 7mo ago
I’m so confused by the focus on “all lawful use.” Yea of course a contract without terms of use implicitly is restricted by laws. But contracts with terms of use are incredibly common, if not almost every single contract ever signed.
130.
by roxolotl 8mo ago
Doesn’t matter whether usage is legal companies are allowed to enter into contracts as they see fit. That’s a core principle of a society with free speech. If Anthropic said you weren’t allowed to use Claude on the toilet they are writing t
131.
by roxolotl 8mo ago
Nice ok I’ll play with that. I’m mostly just learning what’s possible. Qwen 3.5 35b has been great without any customizations but it’s interesting to learn what the options are.
132.
by roxolotl 8mo ago
What method are you using to do that? I’ve been playing with llama.cpp a lot lately and trying to figure out the cleanest options for getting a solid context window on 32gb vram and 64gb system ram.
133.
by roxolotl 8mo ago
Right the red line wasn’t much of a line. If you’re drawing your line only at unconstitutional mass surveillance and allowing the DoD to build skynet because Claude’s not ready for it yet that’s not really a line of principle.
134.
by roxolotl 8mo ago
Yea strong odds this goes to court, the DoD’s clearly inconsistent logic is ridiculed by a judge, the designation is dropped, and everyone quietly goes about their way with the DoD continuing to use Claude according to the existing terms of
135.
by roxolotl 8mo ago
The United States has freedom of speech. The Supreme Court has ruled that money is speech. A company can always direct their money, speech, however they like with regards to the government. Can you be sued for breach of contract? Sure. Is i
136.
by roxolotl 8mo ago
This seems totally reasonable. Even if we say AI can play games and watch movies and have opinions about them, which I don’t think they can, their opinions aren’t super relevant to humans.
137.
by roxolotl 8mo ago
The US has a strong history of trying to avoid building domestic surveillance and a national police. Largely it’s due to the 4th amendment and questions about constitutionality. Obviously that’s going questionably well but historically that
138.
by roxolotl 8mo ago
This is a reminder that we do not need super intelligence to risk catastrophic outcomes. All that we’ve needed is someone willing to delegate the decision to kill to a machine.
139.
by roxolotl 8mo ago
I think the reality is that the maker movement slowed down not because it’s hard to learn c++ but because people don’t care enough. Will maybe twice as many people participate now? Sure. But that’ll still be a small fraction of people.
140.
by roxolotl 8mo ago
So I’ve made very similar comments in the past. This isn’t new information or news. But that doesn’t mean it’s not important to continue to tell people. 3 years ago the state of the art security researchers were pounding the drum on “never
141.
by roxolotl 8mo ago
I think we’re agreeing and likely have different understandings of what’s meant by “code wasn’t the hard part”. What you’re describing is what I’m calling the hardest part: building a system. To me that’s different from “coding”. This isn’t
142.
by roxolotl 8mo ago
People are missing the forest through the trees. Producing code which implements business logic has always been easy. This has always been true, even decades ago. As of at the early 2000s you could have spent around a rather small amount of
143.
by roxolotl 8mo ago
The finding of the first study was people cannot judge their performance with these tools. So I don’t think the lack of individuals not willing to work without them is indicative of productivity improvements. I think it’s indicative of them
144.
by roxolotl 8mo ago
It’s an article about data centers so we’re talking about data centers. 100% agree we should be pushing all industries to use their resources not those of the commons. Data centers do happen to be easier to mostly close loop though than alf
145.
by roxolotl 8mo ago
I’m sure I’ll get flak for this but I’m sorry we can argue about AI all day and there’s a lot of stuff that’s being built and is genuinely depressing but this takes the cake. There’s few memories I cherish more of childhood than being read
146.
by roxolotl 8mo ago
I wonder what something like rabbitmq could look like for this. Agents could subscribe to chosen topics. A topic per agent and then topics per relevant topic.
147.
by roxolotl 8mo ago
This is a great example of how silly this whole thing is. There’s next to nothing to these claws. Turns out that if you give an llm the ability to call APIs they will.
148.
by roxolotl 8mo ago
Here’s some that seem genuinely useful to me depending on what you’re looking for. - https://concrete.style/ - water.css https://watercss.kognise.dev/ - magick.css https://css.winterveil.net&#x2F
149.
by roxolotl 8mo ago
Depends on the car wash. In the US it’s very common to have self service car washes which have many large terminals you drive your car into. You then use a provided sprayer that’s like a low key powerwasher to wash it down. Many people brin
150.
by roxolotl 8mo ago
To call something AI it’s very reasonable to assume it’ll be actually intelligent and respond to trick questions successfully by either getting that it’s a joke/trick or by clarifying.

More ›