sailingparrot

7 ms·

1.
by sailingparrot 27d ago
> The issue with high-profile half-baked interventions is that they distract from real unsexy solutions like grab the plastic in rivers BEFORE they make it to the ocean. You mean they distract from doing… exactly what they are doing? May
2.
by sailingparrot 1mo ago
No, all coast guards announcements are on 16 regardless of severity, because no one is monitoring other channels so any announcements on other channels would be like screaming in the void. The initial call from help from the stranded boat w
3.
by sailingparrot 1mo ago
> The crew are saying that the request came in on a frequency they weren't monitoring There is only one channel you are supposed to monitor, it’s 16. That’s also where the coast guards make all their announcement/requests for h
4.
by sailingparrot 1mo ago
> That isn't cheating. That's a tool doing what tools do. I should have known I was arguing with a bot all along
5.
by sailingparrot 1mo ago
It matters if an article was written by an expert or someone that only has surface knowledge about the topic. It matters if it’s written by someone talented with words or by someone that has to churn 10 articles a day for a content farm. So
6.
by sailingparrot 1mo ago
Is this what AI psychosis looks like ? What are LLMs going to do? Flap the antenna at a specific frequency to achieve faster than light travel?
7.
by sailingparrot 1mo ago
> Isn't this the opposite of what everyone is saying should happen? I think "everyone" is doing heavy lifting here. It's not clear to me at all that a powerful model released with no restrictions would be a net positi
8.
by sailingparrot 1mo ago
There are different dimensions to alignment, refusing to execute offensive cybersecurity actions is part of the alignment stack (that was relaxed here on purpose). Whether a model hacking some infra X when tasked to find a way to hack Y wit
9.
by sailingparrot 1mo ago
I don't have the same read as you. They mention how the offending model is one that had "relaxed" alignment on cybersecurity, on purpose, to evaluate it's capabilities and was never meant to be released. So un-alignment
10.
by sailingparrot 1mo ago
There is no way a Chief Scientist (essentially IC) role is more important than leading the whole of DeepMind, 6000 people strong, working on the most important product for the future of Alphabet. This means Demis wanted a change and to work
11.
by sailingparrot 1mo ago
> DM was sitting on a ChatGPT product for a whole year before ChatGPT got released (LMChat) and Google didn’t let them release it. Heard this multiple time, to me this is pure history rewriting and post-rationalization. OpenAI also had a
12.
by sailingparrot 2mo ago
You are comparing apple and oranges, it makes no sense. > Soccer is by far the most popular sport on the planet, the fact that it makes a lot less money than hockey is insane. Soccer makes much much more money than the NHL, it just doesn
13.
by sailingparrot 2mo ago
> but why do others take their hobo shouting on the street ramblings seriously, presumably only because they have wealth? The other hobos on the street seem to be happily ignored. The wealth is precisely what allows those hobos to act on
14.
by sailingparrot 2mo ago
> Nobody of sound mind wastes time writing stupid comments on the internet if they have an actual motive. I mean Musk? You even used it as an example of "just trolling" in your previous message. He invested north of $300M+ in t
15.
by sailingparrot 2mo ago
> Care about what? About what you yourself said? "Canada could use a system to encourage the wealthy to claim income, where that is incentivized by being able to vote". The ultra-wealthy can already influence elections in very
16.
by sailingparrot 2mo ago
Why would they care? The ultra-wealthy impact election through donations, which is many order of magnitudes more impactful than their personal ballot, even if you were to give them 10x weight. Elon Musk impact on the U.S. election wasn&#x27
17.
by sailingparrot 2mo ago
> hello Wow, clearly Gemini
18.
by sailingparrot 2mo ago
Sure, I'm not saying it's an untractable problem. But it's not as dumb as "how come a team of engineers paid millions haven't even heard of airgapping". Creating a fake internet-like environment good enough to
19.
by sailingparrot 2mo ago
> For security?? Really? By doing that you increase your security during eval and drastically lower your security at prod time, where internet is accessible, and where you are running much much much more requests in parallel, making it m
20.
by sailingparrot 2mo ago
> they also clearly failed to contract specialists like myself to advise them on how to airgap software properly Why would they want to airgap it though? They are trying to evaluate the model capabilities, alignment, potency etc. A model
21.
by sailingparrot 2mo ago
> There is no K4 without Fable 6, GPT-6. That, matters. That's simply not true though. Chinese labs very clearly have the entire stack developed and working. Using traces from claude allows them to shorten their training time by som
22.
by sailingparrot 2mo ago
You have it backwards IMHO, obviously can't prove it, but I would bet that Opus was used to bootstrap Fable. It becomes very confusing since we have started calling everything distillation, but most likely what both Anthropic did for F
23.
by sailingparrot 2mo ago
> So AI will largely hit a wall, frozen at the current level and work will all shift to cheaper inference. Looks like an ideal outcome to me until (if) we are able to solve the alignment issue.
24.
by sailingparrot 2mo ago
Yes they distill, but if you think you can trivially get a frontier-level model by "just" distilling from Claude's public API. you fundamentally do not understand the amount of work that goes into a modern post-training stack
25.
by sailingparrot 2mo ago
> "The yoke of human existence is oppressive" Legitimately go get help and spend more time outside.
26.
by sailingparrot 2mo ago
> When you have a couple hundred billion dollars on the line I have zero faith in the messenger The issue with your reasoning, is that if/when an advanced AI goes rogue, it will necessarily come from a lab with a couple hundred bill
27.
by sailingparrot 2mo ago
If you hate the human condition, you have an easy way out. Why force everyone else to come with you? Is this what depression mixed with the complete unability to wrap your head around the fact that other people might be able to enjoy their
28.
by sailingparrot 2mo ago
> it just was 'paperclip maximizing' to literally exploit Why “just”? Paperclip maximizing is exactly one of the nightmare scenarios. I’m not sure why you take comfort in knowing that it was just that.
29.
by sailingparrot 2mo ago
Hard disagree. You have to assume security guardrails can be by bypassed or will fail to detect an attacker. So if you are going to deploy this model in production non-airgapped, you better know how it will behave in this non-airgapped envi
30.
by sailingparrot 2mo ago
Sure, I just find using OSS grand standing, and trying to educate the audience about the benefits of OSS a bit distasteful after Meta pulled away from OSS for their main flagship models.

More ›