Matthew Hutson

AI might find new physics where we haven’t looked. My latest for @IEEESpectrum: spectrum.ieee.org/particle-physi…

@elonmusk Congrats, but the idea of a Pareto frontier is that there can be more than one "#1." This one has two.

Grok 4.5 is Pareto #1 when considering speed & cost

When people aren’t skilled enough to know how unskilled they are: I discussed the Dunning-Kruger effect with @daviddunning6 for @NautilusMag: nautil.us/why-we-think-w…

Was this @nytimes metaphor written by AI? "As with a mail distribution hub, putting computing power closer to users enables A.I. models to think longer, respond more quickly and complete more complex tasks." Data-center proximity doesn't affect a model's ability to think longer or do more complex tasks. And what does it mean for a mail hub to think longer? nytimes.com/interactive/20…

The World Cup final crowd cam cut from Will Ferrell to Pharrell Williams. A+

@JamesSurowiecki Wait until she hears whom Shakespeare cast to play women.

No - that's an actor playing an ancient Greek warrior.

@heatherr0ads My first thought: He slept a lot and is trying to run it off to tire himself out so he can get to sleep tonight.

roommate just told me he’s “running off seven hours of sleep”

@F_Edits Thanks. It could mean something like “this is so incredible that I need to double-check.”

Great questions - i'm no linguistic expert, but it's basically a method used to humorously emphasise something - similar to a "dramatic pause" (which i guess is what i'd write if i had to use other words) It works well in written format (hard to replica orally; you'd have to do some actually physical theatrics)

Fascinating tactic from Tuchel to tell your team to fall back and invite pressure when your facing *check notes* Lionel Messi

@shatterdvisage @F_Edits Thanks. The sentence started with “Fascinating,” suggesting that one knew the detail was extreme before checking, so I wondered if "*checks notes*" might have another meaning here.

@SilverJacket @F_Edits If you’re serious, it’s a (tired) Twitter joke format that adds an ironic element in that you need to consult some mundane detail which turns out to be an extreme example

@weather_buffalo I don't think the smoke in 2023 was fuchsia.

The latest HRRR shows some absolutely insane smoke on Thursday across WNY. If it comes to fruition it would be much worse than what we experienced a few years ago in 2023.

@Simi_2210_ 19

@Mario_Sneh I'd love to see the prompt you wrote for this story, in case it has more raw glimpses of feelings.

I grew up convinced my dad had a favorite. Every Saturday morning, he’d take my younger brother out for breakfast. Just the two of them. It became such a routine that eventually nobody questioned it. I never asked why. Partly because I didn’t want to sound jealous. The other part was that I was afraid I already knew the answer. If I was being honest, it hurt. Not enough to start an argument, just enough to quietly notice it every weekend. Then one Saturday, my brother came down with the flu. My dad knocked on my bedroom door instead. “Want to come with me?” I almost said no. Not because I didn’t want breakfast. Because I didn’t want to feel like I was only getting an invitation because someone else couldn’t go.

@meghan_daum YouTube

There should be a dating app for middle-aged singles just seeking someone who will pick them up from their colonoscopy. Plus, imagine the name possibilities.

@mcarbin @NaveenGRao @unconvAI Thanks. If you move to physical systems, will you still need to train in simulation?

@SilverJacket @NaveenGRao @unconvAI We train Un-0 with plain ol' backprop

🧠 Today we introduce Un-0 from @unconvAI : the first large-scale generative model build on physics as a compute primitive. This represents a “hello world” moment for physics-based models. We use the inherent time-varying behavior of physical systems to do compute for us. The result is a new way to build a computer that can be VASTLY more power efficient. 🧵 unconv.ai/blog/introduci…

Can LLMs reach "AGI," or do we need world models? My latest news feature for @ScienceMagazine (link in reply). Thanks to Jared Kaplan, @ylecun, @PimDeWitte, @AdamJelley2, @janexwang, @shlomifruchter, @LakeBrenden, @yukez, @merettm, @AnthropicAI, @amilabs, @gen_intuition, @GoogleDeepMind, @OpenAI, etc.

@NaveenGRao @unconvAI In a physical system, how would you tune the parameters, especially the coupling strengths?

@farzyness Would they ever build lithography machines?

Musk Industry is perfectly positioned to serve the best intelligence to cost ratio in the digital and physical world. They are building their own data centers - in space and on the ground. They are building their own chips with Terafab. They are building their own models with Grok, FSD, and the Optimus Bot model. They are building the robots in-house, with their cars and the humanoids. They are building their own power generating assets with solar panels & batteries. They have a massive marketing arm in X. What all of this means is that Musk Industry will have hyper-optimized models married with hyper-optimized chips, running on hyper-optimized robots in the form of cars and robots (and probably drones, ships, and planes eventually), all powered and trained by hyper-optimized data centers in space and on the ground. All of this requires STUPID amounts of money to get started, but once the flywheel gets going, it'll be an unstoppable force. $TSLA $SPCX

Me emociona que ni camine ya, paradito nomás.