NoUsernameGiven
35 posts
SUPERGIRL pre-sales outside America don’t look good and it is tracking to a soft debut at the international box office.
It’s now heading to a $27-$39M debut overseas this coming weekend.
English
@theo Would be cool if you made videos on how you use codex computer-use. I feel like there’s not enough real content about what it can actually do, and real-world applications of it like this. Would be great as it’s underexplored imo
English
I know I dunk on Codex a lot but man this is actually magic.
Theo - t3.gg@theo
Weirdest Amazon Prime day to date. Front page is full of absolute slop. Had to spend hours digging and filtering to find the best deals (thank you Codex) Dropping best deals I found here for y'all. I personally bought or already own every product I share here.
English
changed my profile picture so it's gone for the time being
GIF
English
Why the hell do i have a blue check now. I didn't buy this bullshit
English
@DiscussingFish Fuck yeah time to spread disinformation even more effectively
English
@AndrewRousso Reminds me of @DanManCarney 's great but short lived run club series @RunClubTheShow/videos" target="_blank" rel="nofollow noopener">youtube.com/@RunClubTheSho…
English
@thsottiaux Windows app stability and performance. It hangs, stutters, has bugs, computer use doesn't work (sandbox issues), model selector is slow, opening new threads is slow, etc. Good product, but not a good experience on windows
English
@mattpocockuk Haven't tried 5.2 but used to main GLM 4.7 via the ZAI coding plan and Claude Code as the harness.
Used to be 3$/m, now it's ~16 so the value compared to Claude/Codex is null. I'm sure other providers are much better and cheaper nowadays tho (and other harnesses too). OpenCode?
English
Folks who are running GLM 5.2, how are you doing it?
What harness/provider are you using?
Getting FOMO about an open weights model for the first time
English
@scaling01 Really like your AI takes and insights.
Thought about a newsletter or blog? I can’t quite keep up with some of your more technical things such as the compute/scaling restraints without the context (eg you often quote old tweets and old news but without explaining the context)
English
Anthropic is genuinely just multiple months ahead of everyone
since learning about OpenAI's scaling skill issue, I no longer expect them to be ahead by end-of-year by ~1-2 months
best case for them is catching up to Anthropic or staying at their roughly 2 month lag
Lisan al Gaib@scaling01
SpaceX acquired Cursor I expect SpaceX AI to be between Google and OpenAI by end of year Composer 2.0 was a very strong model for only 1T params, but Elon isn't stopping at 1T models
English
@slaverefugee @scaling01 What? lol. They rejected DoD’s policies, maintained their position and were replaced by OAI. Anthropic got a legal victory regarding the block. There’s been no begging. Not sure what you’ve been reading
English
@fetthasfriends @scaling01 Antrophic didn’t stand up to anyone. They tried to call the shots, got slapped in the face, and came back to beg
English
This should also update your views on certain people on this platform who were saying Anthropic orchestrated this for some press or regulatory capture or whatever
Lisan al Gaib@scaling01
To no surprise it just comes down to pettiness Anthropic is innocent
English
@Nicholi444 Could you post the channel link here or DM it to me? can’t find it
English
@fetthasfriends Exactly the noise around Anthropic being “clueless” misses the bigger picture. The government’s petty streak toward non-bootlickers is the real issue. For sharper takes, Lisan’s telegram channel SCALINGCALLS keeps cutting through the nonsense.
English
@slaverefugee @scaling01 There’s a difference between “handling politics” and grovel/bootlicking. Brockman and OAI are Trump’s top donor and bent to his will the moment Anthropic stood their ground against the DoD.
Dario and Anthropic not bootlicking ≠ everything’s their fault
English
@scaling01 Doesn’t change anything.
It’s just rambling from sources who think they know what’s happening
Still Antrophis fault, everythign is. Fault of the CEO unable to handle politics like ever other CEO
English
@RobinDinVienna @ItsPaddles @xisumavoid Mythos “with protein design and bioinformatics tools but no human assistance, matches or beats skilled human operators”
“conducted novel genomics research in over a week of largely autonomous work..outperformed a recent model published in the journal Science”
+ now code is good
English
@fetthasfriends @ItsPaddles @xisumavoid GenAI is thouroughly useless in medical research. What is used there is machine learning and pattern recognition and that since more than 20 years. In software development, it is only useful for making slop code and having no one know what it even does. I am yet to find a usecase
English
For anyone who cares, GenAI was used on my last VOD thumbnail. I wasn't sure how I felt about it but I also didn't ask for it to be removed either.
Thanks to everyone who discussed its usage. My thoughts on GenAI has been changing recently and I've decided we won't use it again.
English
@Ceborg1 @lovefleki @xisumavoid Talking ab genAI. Mythos “with protein design and bioinformatics tools but no human assistance, matches or beats skilled human operators”
“..conducted novel genomics research in over a week of largely autonomous work..outperformed a recent model published in the journal Science”
English
English
@fabledintime @Ceborg1 @lovefleki @xisumavoid Talking ab genAI. Mythos “with protein design and bioinformatics tools but no human assistance, matches or beats skilled human operators”
“..conducted novel genomics research in over a week of largely autonomous work..outperformed a recent model published in the journal Science”
English
@fetthasfriends @Ceborg1 @lovefleki @xisumavoid That's not Generative AI. You're confusing pattern recognition AI with GenAI
English
@fabledintime @ItsPaddles @xisumavoid Fair but at the speed at which it’s improving it will eclipse human SWEs soon. Not that I believe genAI can run autonomously yet, but it’s gotten to the point where it can definitely 10x a goood engineer. Mostly talking 5.5 pro and Fable 5
English
@fetthasfriends @ItsPaddles @xisumavoid Adding on as someone who does software development, Generative AI currently just does not have the capabilities to "replace software devs" like a lot of companies think. It takes an extra two devs to fix the mistakes a GenAI makes
English
@Ceborg1 @lovefleki @xisumavoid It’s quite good at drug discovery and genomics research, obviously nothing too world changing yet but it’s getting there!
English
English
@ItsPaddles @xisumavoid Agreed for art and writing and such, but do you also think it’s harmful for software development and medical research? Curious
English
@xisumavoid Its good that your point of view is moving, generative AI is incredibly harmful and has no place in any space - especially one as creative as Hermitcraft and its members
English
@lovefleki @xisumavoid Agreed for art and writing and such, but do you also think it’s harmful for software development and medical research?
English
@xisumavoid Please do further research, and speak our against it! GenAI is harming us all.
English
