Tossrock

I feel like Disney's rendering engine already took this name.

Does this mean they'll be able to take the freeways to get there? Surface streets from SF to SFO would be pretty slow.

Altman doesn't have any stock. He's playing a game at a level people caught up on "capitalism bad" can't even conceptualize.

Weird that there was no mention of paramagnetism/diamagnetism in here.

Are you sure you're using 4? Mine says January 2025: https://claude.ai/share/9d544e4c-253e-4d61-bdad-b5dd1c2f1a63

If you were disappointed in the brevity of the article, you may enjoy this multi hour podcast on the ELT, in the form of an interview between one of the scientists working on it and a German CS PhD / podcaster: https://omegataupodcast.net/150-the-european-extremely-large...

They get into the details of the active & adaptive optics, flagship science missions, engineering tradeoffs, etc. The podcast itself is also free and non commercial, so no ads or sponsors!

Actually I think they're tiny homunculi, trapped in a room full of meaningless symbols but given rules on how to manipulate them.

My favorite part of the "stochastic parrot" discourse was all the people repeating it without truly understanding what they were talking about.

Also interesting that GPT4.5 does the best, and also betrays close to the least. Real statesman stuff, there.

I suspect the suggestion of letting them choose names at the start would improve this confusion - the tokenization and tracking of "P#" is no doubt harder to manage, especially for weaker models, than a more semantically meaningful identifier.

The fact that it scales sub linearly with hardware is well known and in fact foundational to the scaling laws on which modern LLMs are built, ie performance scales remarkably closely to log(compute+data), over many orders of magnitude.

That is literally the pitch line for Devin. I recently spoke to the CTO of a small healthtech startup and he was very pro-Devin for small fixes and PRs, and thought he was getting his money worth. Claude Code is a little clunkier but gives better results, and it wouldn't take much effort to hook it up to a Slack interface.

"The Impact Of Taking A Survey About AI On Answers To A Survey About AI" doesn't have the same ring to it.

Spacemonger is this to a tee.

There is evidence that as LLMs increase in scale, their preferences become more coherent, see Hendrycks et al 2025, summarizer at https://www.emergent-values.ai/

Spoiler!

I found this firsthand account of a gig worker trapped by the algorithm pretty compelling: https://zerohplovecraft.substack.com/p/the-gig-economy

It sounds like you're looking for Any Angle pathfinding. The fastest known algorithm for 2D grids is ANYA: https://ojs.aaai.org/index.php/ICAPS/article/view/13609

Some can. Google owns Waymo and runs Streetview, they're collecting massive amounts of spatial data all the time. It would be harder for the MS/OpenAI centaur.

Unsong has a pretty novel take on it.

A funny story: Ambrosia Software was, for many years, THE name in Mac gaming. Their biggest series was the Escape Velocity space sim titles. The final release in the series EV Nova, was popular enough to be released for Windows, as well. Years passed, Ambrosia went the way of the dodo, and now, if you try to run Nova on a modern Mac, it's basically impossible due to all the platform changes since the release... except by running the Windows version, via Wine.

HDI growth in China is levelling out, and irredentist claims to Taiwan are a foundational piece of the CCP's ideology. The "strategic ambiguity" could be tolerated while Taiwan was relatively unimportant, but now that TSMC is one of the most important companies in the world, China is certainly at least considering war.

He actually does suggest a reason in the book, which is a primordial black hole passing through it.

The Basic Factual Information presentation is pretty astonishing. There were so many opportunities to stop and reflect on what a bad idea this whole enterprise was.

During the prototype phase:

"June 2019: OceanGate pilot visually identified a large crack on the internal surface of the carbon fiber hull."

"October 2019: TITAN tested at Deep Ocean Test Facility. Hull showed signs of fatigue, hull derated to 3000M."

On the first expedition:

"Conducted 1 test dive (dive #63) to depth on TITANIC (3,840M) before carrying paying passengers/ “mission specialists"

• 70 equipment issues requiring correction including:

• forward dome fell off during recovery,

• multiple drop weight issues to include the tray being jettisoned due to malfunction,

• drop weight and thruster failure at 3,500M, and

• platform damaged on recovery."

On the last expedition:

"Mission #2 (May 20 – May 28)

• POLAR PRINCE departed St. Johns on May 20th enroute TITANIC with TITAN in tow at an average speed of 7.4 knots (8.4 mph). May 22nd, conducted unmanned dive to 8 meters (Dive 84). On May 24th, after a night of high seas and fog, the TITAN and its platform were found partially sunk in the morning with the tail cone fairing ripped off. May 27th conducted post incident test dive, recorded 13 equipment issued requiring correction."

And maybe the worst, on the penultimate mission before its total constructive loss:

"On June 12th, while conducting Dive# 87, TITAN experienced a variable ballast tank issue, upon resurfacing, an error caused platform to become inverted to ~45 degrees with the bow of the TITAN up, slamming all 5 persons to the aft of the submersible. The TITAN became partially disconnected to the LARS and with the approximate 6 foot [seas?], slammed the submersible and the occupants for ~1 hour until the platform was corrected."

Spending an hour being slammed around in an oversized washing machine really should have given someone time to consider whether maybe their reach was exceeding their grasp. Unbelievable hubris.

Roblox is not a game, and indeed makes its money by giving its engine to developers for free, then taking a cut of their earnings.

The killer app is replacing 80% of white collar employees in 5 to 10 years.

If you say so. I encourage you to invest your beliefs. But it's worth remembering that millions of people are already paying OpenAI and others for the value their models provide. GPT-4 can write code at least as well as a new college grad.

Also, there's no such thing as "Azure credits" at this scale. What they give OpenAI is control over billions of dollars of their capex, which is invested in datacenter scale GPU clusters designed by OpenAI, for their own exclusive use. It has as much in common with a startup getting a $1000 cloud credit coupon as a paper airplane does with an A380.

He will direct Microsoft to continue investing in them? He's the CEO, and their investment in OpenAI has already paid dividends in the MSFT stock price. There is absolutely no world in which he voluntarily decides, "Actually, a few billion is too expensive for my 3 trillion dollar company, time to jettison the biggest growth driver of the past three years."

If you think Nadella will let OpenAI run out of money, you're dreaming.