00:04:52

Claude review of the paper: 00:04:52

https://paste.debian.net/hidden/dde67217 00:34:27 Have people come to some consensus on what should be done with regards to findings detailed in the previous paper? 00:34:43 The one by the same authors (I think) that had to do with Tor 09:46:43 They did report their findings on HackerOne in advance, but when asked to test a patch in their simulation they never followed up. 13:08:43 is the patch already in the code? 13:10:35 not yet 13:10:58 it's not PRed yet 13:11:30 ah ok. i was gonna instruct my claude code to go through the paper and then reproduce the sim in monerosim why not 13:14:36 well still gonna do that. but could also test the patch 13:15:44 dangit got flagged. opus 4.8 it is 17:26:36 @gingeropolous: U tried fable and expected it not to be? 🙃 17:27:17 Fable will flag if you ask it if a chefs knife should be put in the dishwasher 17:31:53 Grok build is more useful than Claude for security 19:50:04 Chat GPT is not bad recently 19:50:27 AI in general are not good 19:53:19 if you have that attitude then you probably arent testing the latest models or have much of an understanding if thats true or not 19:54:15 AI is subject to random errors, which it makes a substantial quantity of 19:56:56 Yes we all no those humans are error free. 19:56:56 it hallucinate a lot, even today 19:57:08 but you can’t just blindly trust AI 19:57:22 Sadly, many do 19:57:26 @slowbeardigger:matrix.org: Apparently you can! 19:57:34 That’s why you must review, read, check, double check etc, and know when is best to use the AI, just a tool after all 19:57:40 @isaac_newton:matrix.org: well yeah but 19:57:44 If you got a head, use it xD 19:58:10 If youre using ai with the thought that "this is a perfect tool which makes no mistakes and I wont have to use any effort" you arent having a realistic outlook on what AI is or how to use it 19:58:10 I was one of the ones who blindly trusted AI spit time ago 19:58:30 @slowbeardigger:matrix.org: Then well, reality + experience + not being fucking stupid 19:58:38 I could read 100 coding books, how long would that take me? I can start working in the same night with those 100 books with AI. 19:58:41 :p you can do decent stuff with clankers 19:58:43 And AI is the only way to do that. 19:58:52 @isaac_newton:matrix.org: Rigthio 19:59:22 Yeah, good luck doing that with any codebase of significant size and complexity 19:59:28 For some people, their only chance of becoming a coder or researcher, might be with assistance. I see AI more as a butler or organizer. 19:59:49 tusko: Thats where learning how to use AI comes in handy, which if you have a hatred for it you are starting off late. 19:59:52 People who use AI to write their code have very little chance of achieving that goal 20:00:16 tusko: Is that from the CIA factbook? 20:00:29 You tell me misses officer 20:00:30 tusko: that’s why you need to know when is best to use a clanker 20:00:46 do I give off feminine energy? 20:00:53 lmao 20:01:10 i dont think there are any bees in this chat, dont worry. 20:01:54 we should probably be using the longue for this type of conversation 20:01:58 the thing is, AI is ok to use, if you dont use that, also fine. 20:01:58 a fucking tool 20:01:58 there was a time that VSC had no autocomplete and basic shit 20:03:01 There was a time when using VSC was retarded 20:03:29 its like calculators. Remember how much they used to seethe about how we wouldn't always have a calculator available. Now its pretty acceptable. 20:03:29 One of my co workers reminisces about how you used to get your fingers hit with a ruler if you counted on them. Sadism is only for the enjoyment of the abuser, it doesn't benefit the victim at all. 20:04:09 Just don’t stop using your brain. 20:05:07 @slowbeardigger:matrix.org: Trust me that wont happen, but I think some people would prefer if I stopped using my brain. 20:05:15 Its usually very obvious after a bit if someone heavily relies on AI for their code 20:05:34 research often requires testing. Sometimes reality testing. :) 20:05:35 tusko: true 20:05:43 Human brain is good at detecting the types of inconsistencies that prevail in such a regime 20:06:08 tusko: What if the AI is being instructed to act with best practice based on actual coding books 20:06:15 tusko: also true 20:06:20 @isaac_newton:matrix.org: you will notice 20:06:24 Still it will fail. I could give you an example. 20:06:33 brain automatically says this shit is not human generated 20:06:54 Like random articles or shit with em dashes 20:06:55 Or they don't follow common software development practices 20:07:17 And then you ask "would someone who wrote something like this really make a mistake like that?" 20:07:27 tusko: there is code that sometimes is difficult to understand or read, not optimal for “human” just to say it 20:07:28 I dont think it matters what the code looks like and in the future people wont care if you use AI to code, it will probably become expected. 20:07:28 Like trying to build a house with your bare hands is considered foolish. 20:07:28 Yeah, you can really tell when industrial materials are used to build homes. You're right. 20:07:47 The thing is, no one will care. 20:08:07 yeah and the tool will get good enough 20:08:15 I think quite the opposite might occur. AI makes random errors. Other AI train on material containing random errors and makes other random errors. ... compunding errors 20:08:23 not to blindly trust it but the quality gonna be good 20:08:40 ive been using AI for a number of years a bit too much, trying to use it to code, and I can say that its currently in a state where it is good enough. 20:08:41 We will likely have semantic errors all throughout codebases before too long 20:09:03 We learn it learns 20:09:04 @slowbeardigger:matrix.org: obviously blind trust never works, the AI needs more and more context only a human can give 20:09:15 AI doesn't learn 20:09:26 learning is a human activity 20:09:31 you're not going to build a massive application by sending one intern to go get coffee, thats the equivalent of thinking one prompt will build your app 20:09:37 tusko: objectively false 20:10:01 Even if saying today that AI doesn’t learn is somehow valid 20:10:01 just a matter of short time 20:10:50 I guess you can get really pedantic, AI condenses information. and it needs input to do so. So whatever it is, resembles learning and grows like learning does 20:11:04 I just fundamentally disagree but we're in some kind of philosophical question now 20:11:41 I can cede that you can make philosophical arguments about "what is thinking" "what is learning" etc and thats a deep rabbit hole 20:12:56 I could make a number of philosophical arguments, ill try to avoid that here. 20:30:59 please take it to lounge, not a lab conversation