522 Photos and videos
Normally I wouldn't say anything at this stage but I feel that the moment calls for it. This isn't directed at any particular company (I'm serious), but it's something we should all start thinking about now.
I'm moving to NYC to join @pangram
The midterms are coming up, and I'm worried about AI generated misinformation and other misuses of AI. Even right now the bots are on overdrive bc of the war, this is bad.
My old job paid ~210k/yr, more money than I knew what to do with. (that's also why no new blogs/research)
I don't expect to stay at pangram long enough to vest, and so in all likelihood I will be taking a 40k/yr pay cut to do this. And of that income I intend to donate a good amount (political) too. I will be leaving all my friends here. I've gotten vastly better at making new ones but is still very uncomfortable. And I will be going in person instead of working online. (None of this is pangram's fault btw I intentionally didn't negotiate, and they're aware of all the details)
My questions to you are:
1. How much is your conscience worth?
2. How easily can your conscience bend?
3. Will ambiguity, intentional or not, satisfy you?
4. Will you simultaneously be capable of good compromise when the time demands it?
5. Can you, right now, in good faith, say that what you are doing is right?
Our society is falling apart at the seams, trust is degrading, loneliness is at an all time high, antisocial behavior is common, gambling is everywhere, influencers sell out the truth and their viewers for money.
We are building an extraordinarily powerful technology while it all happens.
Are you prepared to be the people the world needs right now?
21
9
217
20,436
ueaj retweeted
Jun 13
An Anthropic hype machine believer dies and goes to heaven. Upon arriving, God grants them the chance to ask one question about anything in the universe. The hype believer eagerly asks, “Did the Trump administration ban Mythos to hype Anthropic?”
God replies: “The Trump administration believes Mythos poses a real national security threat, and they acted alone.”
The theorist pauses, nods slowly, and mutters to himself, “Wow… this goes higher than I thought.”
Jun 13
the anthropic hype machine strikes again
8
39
917
60,123
Okay, but in retrospect, the world would probably be a better place rn if Steve Jobs had come out and said:
“This revolutionary device will completely erode social norms, tank TFR, and monopolize your attention for 14 hours a day at the expense of human interaction.”
Jun 13
If Dario had invented the iPhone he would have spent the whole keynote talking about how it might blow up in your pocket
13
49
808
31,483
ueaj retweeted
Jun 13
GPT 5.6 sandbagging evals to dodge export controls
12
80
1,714
88,043
> Elon is wealthy and powerful not strictly because he is greedy, but because he also truly wants to colonize space. That too will require immense power.
Apropos of nothing
We don't get to choose whether the people at the top are power-hungry, that much is inevitable. But not all power hungry people are bad.
We don't typically think of them this way, but, activists are some of the most power hungry people in the world. MLK's dream requires an immense amount of power. The founders of this country needed an immense amount of power. They all acquired it in different ways, but they were all power-hungry.
And ykw? Good for them.
*Will to power* and competence is what determines who gets to the top. Greed / sociopathy is only one such will to power. but there are many others. Elon is wealthy and powerful not strictly because he is greedy, but because he also truly wants to colonize space. That too will require immense power. Anthropic is similar, they're not doing it for the money.
Good for them.
1
8
528
The real reason synthetic data is bad is not because it's fundamentally wrong, there are principles for doing it correctly, but because some of your company consists of people like this
Jun 12
An underrated part of this discussion is that
(a) there's huge leverage in improving data, and
(b) there's no way Anthropic could safeguard this
xAI could instruct Fable to look through EVERY row of pretraining data and fix any typos and errors. this probably the single highest-leverage activity for a lab playing catchup
and it's not possible for Anthropic to prevent this without completely kneecapping the model itself, because data quality work looks like any other kind of knowledge work ("check this text for errors", "rewrite this in a formal tone")
6
4
51
7,922
ueaj retweeted
Jun 12
Today, we filed a lawsuit to permanently dismantle a group of organized cybercriminals accused of using AI tools — including Gemini — to scam Americans via fake text campaigns. Here’s what to know:
◾Our suit targets core software developers in a cybercrime operation known as the “Outside Enterprise.” The group has allegedly weaponized AI to quickly generate highly convincing fake government and brand websites intended to steal victims’ credit card numbers and personal information.
◾The group used AI and different Google products — including our trademarks and logos — as part of these phishing campaigns.
◾The scale of the operation is massive: More than 100,000 victims have been scammed, with losses estimated in the millions.
106
329
2,866
370,884
ueaj retweeted
Jun 12
1
40
1,367
81,779
suggestions for bugs in claude desktop. I like the attention to detail in CC and I think it's its biggest advantage but I wish that taste was spent on desktop more as it's clearly the long run better platform.
1. esc doesn't work to exit settings
2. I can't see all concurrent subagents without clicking a fairly hidden single note in the context window
3. I can't talk to the subagent, and neither can claude (this is also a claude code issue)
4. it seems to be a lot less autonomous in claude desktop but maybe that's just me
5. on my 4k monitor it seems a little too zoomed out, straining my eyes, I can zoom in ofc but the default zoom is poorly calibrated
6. ideally we can manually organize subagents or workers in hierarchical structure, even sharing them between claude code sessions or making folders of subagents (and allowing the same folder to belong to multiple folders) so I can manage worktrees, etc.
7. when an agent interrupts the main agent there's nothing in the UI, the model just starts talking, I'd bet this also holds for concurrent bashes and monitors. I'd also bet the bashes/monitors don't have an easily accessible window
8. subagent progress is not visible in the window, it just shows the input prompt and the result when done. Ideally I can see it and even talk to it, or even open it in the main window as mentioned earlier.
9. ideally I have an option to run everything in WSL
10. the message queue display seems buggy, if I enqueue a message it'll sometimes jump up to the currently being processed user message (ahead of claude's response ongoing response, it makes it look like claude already saw the message, which it didn't) (see image)
10. the default should be drop down the "Running xyz" until the turn is complete so I can see it's progress and then it should shrink back (unless the user manually closed and opened it again or opened some sub-window) (image 2)
11. user messages should be right aligned instead of left aligned like the agent
12. blue on blue for user messages has poor contrast bad color
13. pressing escape before claude starts thinking doesn't return the last user message to the bar for editing like in the terminal
14. in settings, and perhaps in other areas, when zoomed in, I can't scroll down, as certain flags are cut off. actually even when not zoomed in some settings are cut off and can't be accessed
I don't think Claude can show me images on desktop, obviously not relevant to terminal but it would be nice if it could do visualizations and show me rendered outputs. IMO the GUI should be the primary focus and not the CLI, it has a much higher ceiling for UX.
@bcherny @trq212
4
9
1,923
@amorriscode @felixrieseberg thank you ❤️ it looks like almost all of these were fixed! Can we also get WSL support? That's the only missing ingredient for moving everything to the app for me.
174
I think it's insane to think nuclear proflieration is ok and that it's a far more extreme position than anything Dario believes. We have put an enormous amount of power into preventing it from spreading, the top down authority of other nuclear powers is what stops us from going extinct.
Basically all my political oomfs are completely fine with what Ant is doing, and I suspect most normal people would be too.
Jun 10
nobody’s launched a nuke since the unipolarity days.
central authorities best serve the interests of the people when their authority isn’t guaranteed.
1
1
11
1,536