nobodywillobsrv

6 ms·

31.
by nobodywillobsrv 8mo ago
Of enemies. Of enemies. There are probably three modes of safety. Deploy tech to unknown group. Could be enemy could be friend. You disable it's abilities perhaps. If you deploy tech to friends you might enable more defense. Anthropic
32.
by nobodywillobsrv 8mo ago
It really feels like I am no longer impressed with Anthropic safety. Do they have even a basic understanding of the different regimes of safety and what allegiance means to ones own state? It would be fine to say they are opting out of all
33.
by nobodywillobsrv 8mo ago
I noticed this a while ago. And the op isn't even experience the degradation of what could have been a huge platform: FB marketplace. I thought during the pandemic FB marketplace was going to go somewhere. I thought they would try to s
34.
by nobodywillobsrv 8mo ago
Isn't visa free travel just going to create more problems at the border where border force has very little time and no paper work to review your case? Pre applying is usually less stressful as you are trying to do your best to show tha
35.
by nobodywillobsrv 8mo ago
To be clear, what I mean is that grok will query 30 pages and then answer your question vaguely or wrongly and then ask for clarification of what it meant and then it goes and requeries everything again ... I can imagine why it might need t
36.
by nobodywillobsrv 8mo ago
Why not just say "reboot machine when done" if that is what you mean? It explains what it means.
37.
by nobodywillobsrv 8mo ago
Exactly. Interesting that both orwell and azimov were wrong in different ways. Also azimov seems unaware of the Fabian link to the title which is surely a factor in the origin.
38.
by nobodywillobsrv 8mo ago
This missed the point that they are ignoring evolution is literally the way you build things. There is no other way. You don't know what is actually needed or what might work really. You try things and then compress later. If you can t
39.
by nobodywillobsrv 8mo ago
Yes I have found that grok for example actually suddenly becomes quite sane when you tell it to stop querying the internet And just rethink the conversation data and answer the question. It's weird, it's like many agents are now i
40.
by nobodywillobsrv 9mo ago
It's worth stating that addictive tech is addictive partly because it doesn't work very well. So they build useful things and then make them pretty bad and less useful. If they were useful your interest or need would complete and
41.
by nobodywillobsrv 9mo ago
I would imagine there are a lot of "small nice to haves" that people submit because they are frustrated about the mere complexity of submitting changes. Minor things that involve a lot of complexity merely in terms of changing som
42.
by nobodywillobsrv 9mo ago
The fixed fee plan is because the agent and the tools have internal choices/planning about cost. If you simply pay for API the only feedback to them that they are being too costly is for you to stop. If you look at tool calls like MCP
43.
by nobodywillobsrv 9mo ago
Anthropic should do a full inquiry into this and fire people for it. No questions asked. Fire them. And then reply with automated bits.
44.
by nobodywillobsrv 9mo ago
How about both worlds? Instead of asking the agent to execute it for you, you ask the agent to write an install.sh based on the install.md? Then you can both audit whatever you want before running or not.
45.
by nobodywillobsrv 9mo ago
AI (chat) companies now have enough data to recommend h2h (human to human) but they are not building this. They could literally find people who are working on he same things and recommend them for networking etc. But that would take you off
46.
by nobodywillobsrv 9mo ago
What exactly was the bad stuff? He was insensitive about empirical reality or he was literally wrong about something in the sense of being very confident about something despite having little data? Or something else? I only remember the car
47.
by nobodywillobsrv 9mo ago
Can you not use PAL MCP for this? Have one top agent as controller etc? It's not ideal but it feels like the space of multi agent stuff is evolving ... I notice that there are a lot of posts on hn about these things so we are trying to
48.
by nobodywillobsrv 9mo ago
It's nice to see someone else going mad, even deeper down the well. I don't known the details but I was wondering why people aren't "just" writing chat venues any commns protocols for the chats? So the fundamental u
49.
by nobodywillobsrv 10mo ago
Thank you for saying this. I often find this "glib" explains of ML stuff very frustrating as a human coming from an Applied Math background. It just makes me feel a bit crazy and alone to see what appears to be a certain kind of p
50.
by nobodywillobsrv 10mo ago
Everyone should be simply posting algorithmic content to Facebook. Screenshots, etc not giving them your own life stuff imo. We need to push back on personalized feeds. Share a high percentage of what you see so that there is a digital comm
51.
by nobodywillobsrv 10mo ago
I would be curious to know if the treatment of statues in terms of "making them ugly and ridiculous to the point of being insulting" is roughly uniform across the different historical cultures being treated to this "reconstru
52.
by nobodywillobsrv 10mo ago
How do you know they don't? How can you show this? It's fine if it's just vibes but just want to know what is known and can be said.
53.
by nobodywillobsrv 11mo ago
Yes, I think it's worth getting details ... like my mental model is that everything within 2x is sort of reasonable error. Looking for 10x errors and cliff edges like in the 2007 crisis where I think a good anecdote is like default pro
54.
by nobodywillobsrv 11mo ago
The piece denies the existence of problems though. It starts of with enragment and denailism. The piece is the polarization. "Diversify your information" in no way cancels single events and single counter examples. The idea that i
55.
by nobodywillobsrv 11mo ago
No, one should always present a path of how to explain it. This is hard. get the data. gather it.
56.
by nobodywillobsrv 11mo ago
There is a lot of bubbliness sure but some of the rhetoric is a bit sloppy. Like "swapping money back and forth" arguments is literally was economies and specializing results in. The debt securitization could be an issue but one t
57.
by nobodywillobsrv 11mo ago
Anything involving what sounds like genetics often gets blocked. It depends on the day really but try doing something with ancestral clusters and diversity restoration and the models can be quite "safety blocked".
58.
by nobodywillobsrv 11mo ago
Interesting that Claude also just hallucinated information as it does for all of us. But perhaps a better guardrail would be not to refuse things like this but frustrate the use by giving them fake results in believable ways. A stupid but h
59.
by nobodywillobsrv 1y ago
The PR spec for some open source projects are quite onerous. What is unspoken here is that some open projects are using cost of submission AND cost of change / contrib as a kind of means of keeping review work down. Nobody is correct h
60.
by nobodywillobsrv 1y ago
How to do simple change on complex projects at the same ROI with LLM?

More ›