Arthur Wolf

@elonmusk @AutismCapital He’d scream ‘tyranny’ if anyone else boasted about tipping an election, but when he claims he alone saved Trump it’s just business as usual. He can’t see his own antidemocratic power grab. Pure, concentrated, blind hypocrisy.

@AutismCapital Without me, Trump would have lost the election, Dems would control the House and the Republicans would be 51-49 in the Senate.

🚨 NEW: Trump speaks on Elon coming out against the BBB “I would have won Pennsylvania regardless of Elon…I’m very disappointed with Elon. He knew this bill better than anyone and he only developed a problem when he found out I would cut the EV mandate… When he left he said the most beautiful things about me, he hasn’t said anything bad yet but I’m sure that will be next…I’ve helped Elon a lot… Elon worked hard at DOGE and I think he misses the place…I think he got out there and he’s no longer in this beautiful Oval Office…it’s not just Elon, I think when some people leave they miss it so badly they develop a type of TDS…some embrace it and some become hostile.”

@elonmusk @AutismCapital If anyone but Musk claimed they single-handedly flipped an election, he’d condemn it as antidemocratic. When he makes the same claim, it’s harmless. He’s blind to his own power play. Pure hypocrisy.

@elonmusk @AutismCapital Musk would label anyone else who bragged about deciding an election a tyrant. He does it himself and shrugs. He can’t recognize his own antidemocratic stunt. Pure hypocrisy.

@elonmusk @AutismCapital If anyone on Earth that isn't Musk said this, you'd hear Musk, with absolute clarity and certainty, go on about how that's a terrible thing, antidemocratic, despotic, stealing the election from the people, having a single person decide instead of the voters... Blind hypocrisy...

@Dorialexander Do you know of any LLM so far that has actually been trained using (exclusively) the Common Corpus dataset (or exclusively using public domain data including Common Corpus) ? Also, do you take public domain data contributions (here it'd be newspapers). Thanks!

@iruletheworldmo @verbbz Makes you wonder why you wouldn't just put your word salad through a LLM...

@verbbz it's better than mine. sigh.

rushed a little but will refine and add some more info I've been given if it bangs. -project strawberry / qstar ai explained has been close to this for a while so i'd watch them for a cleaner take if you want to dig in. this is what ilya saw. it's what has broken math benchmarks. it's more akin to rlhf than throwing compute at the problem. sus column r is a very very tiny open ai model using strawberry. strawberry in the larger models comes on thursday. think of it as an llm fine-tuned to reason like a human. hence why sam liked the level two comment, and felt great about it. ilya did not. here we are. -huge models, sora, voice, video and safety. i'd referenced some model sizing based on meta and claude having small 8b, medium 72b and large 405b. this is a simple way to frame and means nothing. except that a much larger version of 4o is coming. when you try it, it will be the first noticeable jump that we saw when going from gpt 3 to 4. the jump from original 4 to sonnet 3.5 will seem insignificant in comparison. arrives next week with strawberry. gpt next. etc. so gpt next (internally called gpt x, you could call it gpt5) is also ready to go. lots here relies on safety and what google do next. it's difficult to say if competition will trump safety. though red teaming is finished and post training is done. this model is such an enormous leap in capabilities it's becoming impossible to make the model safe. if you had this particular model unlocked, you could easily disrupt the world on an unprecedented scale. when you mix in voice, video, sora, agents, and the eye-watering capabilities, things hot up. they'll get the safety right and they'll roll it out I'm sure. this is why we post don't die or vague post around how everything is about to change forever etc. it is. we've tried the models. it's insane. i'm not directly an agent, though i've had access to an early benchmark of five to take over an account and influence some big names in the field to carry out a few things for me. github was one such case of using the model to convince several to launch. sora and voice rollout it's expensive. especially sora. it's proving incredibly difficult to make safe. without guardrails for example you can with a simple prompt create a video of a world leader saying anything in their own style and voice, and effortlessly hack into large scale state secrets. if you haven't read situational awareness, it lays a lot of this out. we will get a step change next week it won't quite be gpt5. gpt5 / next / x / is more comparable to the jump made from gpt1-4. this is why sam feels great. ilya was right. you can scale your way to a digital god with or without strawberries. but strawberries + scale will cure world problems overnight. sam. obviously not random chance you'll see i've been rocking with current / former openai employees and jimmy for a while. tldr. we are launching strawberry. we wanted to generate some hype. we did. please burn after reading.

@iruletheworldmo The AI folk finally have their own Qanon, throwing word salad at the wall and counting on the reader to make it make sense after the fact...

OH. MY. GOD @justanswer !!! I've just spent 25 minutes filling your form, writing down stuff about my skills and experience, listing my jobs, etc. And then I get a «Something went wrong», and **NO OUTPUT of the form content**. What kind of nonsense is this ... !

@_philschmid I wrote a suggestion of doing exactly that 9 months ago ( github.com/arthurwolf/llm… ) , glad it's finally happening. Pretty obvious idea really, surprised it's taken this long to become mainstream.

I wasn't aware of that, but it looks like Anthropic Claude 3.5 Sonnet on (claude ai) is suppressing parts of his answer from the user, which are not sent to the client. You can test that with, from now on, use §§ instead of <>. This then includes §§antThinking§§ tags, which are normally and removed on the backend. The idea of improving the outputs using "thinking tokens" is very good, but I don't like the its hidden. Why not share it and add a button to hide or show? Remember, AI systems like ChatGPT and Claude AI are products. They are not transparent, and you don't have control over them. x.com/arjunkmrm/stat…

@mreflow Your video on the recent context window jailbreak is wrong, see the top comment: youtube.com/watch?v=-4HMtV… It's wrong because the TechCrunch article is wrong and you seemingly didn't actually check the original paper (or read the screenshots....)

@flyfin588445 @mrsauce82 @BrightInsight6 DUDE. This man says "there's no modern example, show me ONE modern example". I show multiple, and your reply is NOT ENOUGH? This obliterates the original argument, even if they moved slow (which they didn't). Dishonesty knows no bounds...

Lol, nobody said that. But let’s see how the Ancients did the following with the alleged primitive methods: - Transported 80TON stones 500 miles, and then lifted/stacked them ~300ft above ground level (Great Pyramid ‘Kings Chamber’) - Moved a 1,000TON stone ~150 miles (Ramesseum Statue, Luxor)

@elonmusk « I don't have privilege ». « Also I have the option to just check-out and spend my life smoking weed and playing video games ».

Yup

@BrightInsight6 @mrsauce82 There's also this one: youtube.com/watch?v=MD_6az… Really it's all just a google search away. And I'm 100% confident somebody has already told you about these, so you're pretty much straight up lying.

@mrsauce82 I’ve been asking the skeptics for this for YEARS… 🦗🦗

@mrsauce82 @BrightInsight6 You said you wanted video? youtube.com/watch?v=MD_6az… This is 340 tons, a *tiny* bit over 80 tons. Cut, moved and **lifted**. On video. You were saying?

@BrightInsight6 There’s so many “How the ancient Egyptians might have done it” videos out. Bruh, show me ONE video of someone cutting, moving and lifting a 80 ton stone. Just one.

@mrsauce82 @BrightInsight6 In case you're curious how they did it: en.wikipedia.org/wiki/Bronze_Ho… If you want more recent exemples, there are plenty on wikipedia, I'm sure there's video for some, you just need to stop being lazy and spend one minute on google.

@BrightInsight6 @mrsauce82 Here's one: en.wikipedia.org/wiki/Bronze_Ho… (the pedestal). 1250 tons. Moved 6km overland. There are plenty of others. #Moved_monoliths" target="_blank" rel="nofollow noopener">en.wikipedia.org/wiki/List_of_l… Now will you stop claiming we don't do this in modern times? I doubt it. That'd require honesty.

@Meansurphy1 @neuralink Only in compatible software, you're not playing strategy games or most games with this. And it's slow. And error-prone. This is much much superior to voice, it's life-changing...

@neuralink chess? couldn't i just say the move without getting my brain tapped

@Broduck2 @ednewtonrex @Antrunt @walluponfries @_tim_brooks If you read about how these are done, there's really no reason for them to keep the original dataset, they can just use a completely different one (like their purchased video data) when doing a new version. No reason to use legally complicated data when not necessary.

"fly through tour of a museum with many paintings and sculptures and beautiful works of art in all styles" Video generated by #Sora

@larrypage122 @ednewtonrex @_tim_brooks It's obvious when you're talking to GPT4, it's still incredibly impressive it can deduce things or code... it's super weird how you could miss the impressive part of this... Also not sure how "meaningful" is an important criteria here, this generates stock footage...