zmmmmm

6 ms·

1.
by zmmmmm 3d ago
I agree, it is very dangerous that it seems like there is not going to be accountability for these incidents - from either legal or regulatory point of view. In fact, I would say that is the main danger. If someone was in jail right now d
2.
by zmmmmm 5d ago
It seems like all this happened in the same time period earlier this year. It makes me wonder if all of these were part of a single larger incident where multiple experiments were run with insufficient or missing constraints or an unknownin
3.
by zmmmmm 6d ago
This idea of remotely hosting the agent harness is honestly backwards to what I need. In so many cases, all the friction is about how to provision access to local data so the agent can work. So you started with the problem of how do I integ
4.
by zmmmmm 7d ago
While this result does imply there was some training on the reasoning trace and output of GPT 5.5, it doesn't tell us how much of the source of its training it was (even a small amount of post training could bump up the correlations in
5.
by zmmmmm 8d ago
Is this just Google precomputing Alpha genome values - which were already accessible via API and making them available as another API (presumably more broadly)? Or is there actually new information?
6.
by zmmmmm 12d ago
wouldn't it be interesting if nVidia buying hugging face was part of hushing up the fallout there On the face of it, they would have very good cause for some action there, assuming they wanted to.
7.
by zmmmmm 12d ago
> Sorry, I guess we will put up better guardrails next time Or, if you are Anthropic: > This illustrates the risks posed by open models!
8.
by zmmmmm 12d ago
One crucial detail here that differs from the previous incident is this was a vanilla reasoning type task. Even as concerning as it was, I always evaluated the previous incident differently because it was inherently a cyber security /
9.
by zmmmmm 13d ago
Agree. It wasn't as drastic but I think Vue did the 2=>3 update at an unfortunate time. They lost a lot of users along the way but they also confused a lot of the LLM training too, I am sure. Still, I have great success with it, esp
10.
by zmmmmm 13d ago
Vue had a good tenure as a solid #2 to React, so I think it probably has got a lot of representation in training data. Things really splitered after that but it got a good foothold.
11.
by zmmmmm 14d ago
It definitely leaves a bad taste because it is completely transparent their concern is not security here and that means they are lying / misrepresenting this to our faces - which then raises the question of whether you can trust them o
12.
by zmmmmm 14d ago
I'm curious what your methodology is that results in that? Are you running multiple teams of agents all adversarially reviewing each others code? Lots of different projects in parallel? I've only rarely maxed things out and then i
13.
by zmmmmm 15d ago
Obviously Ed is a special case, but let's be honest, pretty much anybody telling you they can predict the future is selling you BS. And that includes all the people confidently predicting he was wrong. The actual situation is, there ar
14.
by zmmmmm 15d ago
not having to implement proper table auditing?
15.
by zmmmmm 17d ago
which is again where Anthropic has outplayed them. Because Claude is being happily applied everywhere as a universal term, with "Code" or "Cowork" only appended as necessary.
16.
by zmmmmm 17d ago
Yeah ... the simplest answer to what ChatGPT Work is seems to be a panic-clone of Claude Cowork as a hail mary to try and catch up to Anthropic in enterprise. Cowork is like crack cocaine to nearly every exec I've seen use it from the
17.
by zmmmmm 18d ago
It's strange to me that there is not a more conscious call out that letting an agent edit its own behavior crosses an explicit risk threshold that requires additional controls. They happily drew the whole loop at the top of the page wi
18.
by zmmmmm 20d ago
The "good enough" concept is interesting because of how systematically people over estimate it. So often, things that are lower quality but thought to be "good enough" turn out to be either not good enough or not worth
19.
by zmmmmm 22d ago
there is an inbetween .... i insist people interactively rebase those commits out. In some contexts it is actually important to have traceability of iterative proof of work towards the final result.
20.
by zmmmmm 22d ago
Moves like this are nearly always a precursor to a user hostile change of some kind. I suppose anybody who didn't leave already are likely so rusted on they still won't budge, but people should definitely look for open alternative
21.
by zmmmmm 22d ago
Anyone dealing with sensitive or regulated data can get started instantly with local models where they may need time consuming process or complexities to send sensitive data outside.
22.
by zmmmmm 23d ago
The big question / risk is whether there are unknown tipping points where things run away. Many are talked about - ice melting, reducing reflectivity, more heating causes more heating etc. But some fraction of these are probably unkno
23.
by zmmmmm 24d ago
The real problem for parallel dev is ensuring parallel dev environments can seamlessly co-exist without treading on each other. As soon as one of them wants to open a port, talk to an external database or write to a shared location as part
24.
by zmmmmm 26d ago
If you initiate a wipe before approaching immigration, which swaps the whole phone contents to an encrypted backup that you physically can't decrypt without a key that (say) a friend knows. Then you would be offering immigration the de
25.
by zmmmmm 26d ago
> Larger images are scaled down while preserving their aspect ratio, so that the total pixel count after resizing is roughly that of an 800×800 image. It's useful but for OCR and a lot of other applications it needs to be a bit high
26.
by zmmmmm 27d ago
Make more TUIs .... The main thing I like is that TUIs are guaranteed to be navigable by keyboard. So many non-TUI apps don't include shortcuts at all for critical actions. It makes the first week of using the app more efficient and th
27.
by zmmmmm 27d ago
It is also the case that the copyright holders are often putting restrictions around use of electronic forms that are driving the desire to use physical copies. I doubt AI companies would use a single physical book if they could avoid it -
28.
by zmmmmm 27d ago
it's sort of non-controversial at the extreme end I think - where there is no human contribution, why should there be protection? But it just begs the actual question of how much human contribution there needs to be: - I wrote the
29.
by zmmmmm 27d ago
but what if I really am a moron? how do I get that level of explanation now!
30.
by zmmmmm 28d ago
energy / time / money / environment / office space reclaimed

More ›