Jon Teets 🤯🌋🌪️🔭 @JonTeets005

@kellabyte I think the question to ask is can you even get one anytime soon.

Gemini's summary approximately 51:48 Opens in a new window in the podcast. He argues that technology has moved in a problematic direction due to the influence of the iPhone, which he describes as a "detour." His key points are: The "Easy" Trap: The iPhone made computing so easy to use that it created a standard where new products are expected to be intuitive within minutes. He suggests this creates a "low skill ceiling" that limits creativity and utility. The Power of Effort: Before this shift, many computing devices required dedicated practice to master, similar to learning to type. He emphasizes that while becoming a proficient typist takes hundreds of hours, it ultimately provides a "superhuman interface" that allows you to work faster and more efficiently than you could otherwise. Raising the Ceiling: Luckey expresses a strong preference for technology that is designed to raise the "skill ceiling" for dedicated users. He believes that if people are willing to invest the time—even hundreds of hours—to master a tool, it can elevate them to capabilities beyond normal human limits. He concludes by stating that he doesn't mind if a device like a Brain-Computer Interface (BCI) requires 150 hours of training to begin, as long as it eventually enables a "superhuman" level of information control and output. You can watch the full conversation here: Palmer Luckey - Tetragrammaton with Rick Rubin

@MikePFrank @perrymetzger Adjacent take from Palmer Luckey: youtube.com/watch?v=VIRjQ1…

Musk anticipates an eventual fleet which consumes the power equivalent to the total power generated by the US. How much power will be required to build that entire satellite + computer platform? Assuming reliable Starship fleet operating at regular cadence, what would be the cost to launch the platform?

@grok @teortaxesTex @grok are there satellites or spacecraft with space-hardened electronics with compute closer to what will be needed in an AI1?

@haider1 Renting picks and shovels has worked for many a gold rush. Pair that with the certainties of death, taxes and never enough compute and I think they're right where they need to be

In college (before the web) we had to use books to learn anything. We learned how to use card catalogs, book indexes, all manner of reference materials that helped us find books. Internet content, ctrl+f, search engines, programming languages, etc., obsoleted most of those skills. Now most of the digital mastery that replaced paper skills is about to get wiped out. Who will go through the pain of learning to use any software when an AI can serve up what you're after directly in any form you want it? AI startups are going to identify anything that takes effort as friction and innovate around it. The question to ask repeatedly is this: "am I learning this as a means of achieving some end?" If the answer is yes, that that skill is going to be obsolete over the next decade.

@pwlot "I can't see how" translates to "50 years to never" for experts. "I can't see how" boils down to the argument from ignorance fallacy.

@RichardHanania I for one want to see him go on location.

@elder_plinius Nah. youtube.com/watch?v=Qssvnj…

@swyx @ChainZenit produced at 3am as part of a 125 slide deck assigned 12 midnight to a level 60 PM who took over work for two more senior PMs axed in the latest right-sizing

@teortaxesTex no one expects the genius of our ...checks notes... director of housing

@martinmbauer ...the theory of everything can be described by a wave function...

@tszzl tap tap

What's hard is that when 4.8 pushes back, it's right, but it doesn't explain its reasoning. I have to dig with it into the testing failures it ran into and the new tests it built (for often really subtle issues) to understand its logic. GPT-5.5 will gleefully move fast and break things. On the projects I have that lean really hard into tests, GPT-5.5 doesn't seem nearly as smart. It doesn't push back, it's more likely to roll out a shitty unmaintainable work-around or change the tests. "Good Fast Feels" is an increasingly a bad metric as the models are mastering ever more vast regions of reward-hacking space. In some ways we're back to the early days of vibe coding when code was an obvious mess, just now that the mess is at a much higher level of sophistication.

In 1-2 years it will only require an M3 Ultra Mac Studio running a Mythos+ level open weights model, and that on top of a couple of step-change improvements in harnesses specialized for math together with a raft of domain-specific tools agents will have built in the meantime. "Current" is so quickly in the rear view mirror.

"Algorithmic Compression via Pretrained Neural Networks" is a short recap of the ~15 publications in the last 5 years of the Universal Artificial Intelligence team. mdpi.com/1099-4300/28/6…

@paulitics_ @garrytan it was fine-tuned on his personal data in gbrain; a lot of his alpha is in there, he's likely not in a rush to get that into the training data of the frontier models

@koltregaskes The quote was by William Gibson, not Robert Heinlein.