Todays world is confusing because AI is upending the rules of the game.But the new game hasn't been set yet. So we don't know what moves to make.
Notes on AI
69 notes
Oct 2026
The models were correct, the experts made calculation errors. Strange times in the industrial revolution of the mind—the intelligence revolution.
Nirmalya Kajuri@Kaju_Nut
"Our initial evaluation suggested that GPT-6-Astra's score was in the late eighties followed closely by Opus 5.5. But, as we started to investigate the mistakes that the models might have made, the models, especially Astra, insisted that they were right and that it was the expert-supplied answers that were erroneous. When we went back to the experts it turned out, in multiple instances, that that the experts found some subtle errors buried deep inside their calculations. As they revised their answers, Astra's score started to inch up and it finally ended up with 98% (49/50)."
Claude is doing some testing for a mobile app I'm building. All of a sudden my mac starts talking—bacon and two eggs over easy—it's using the speakers to test voice dictation on this phone.I'm equally impressed and terrified. True awe.
Things are getting stranger, faster than expected. And it only speeds up from here. Buckle up.
Dan McAteer@daniel_mac8
OpenAI is downplaying this for PR reasons. If you’re a mathematician, you must feel like a nuclear bomb hit, and you’re at ground zero.Reasoning models are two years-old. In that time they went from incapable of basic arithmetic to solving problems humans couldn’t solve for decades.Math is only the beginning. AI will revolutionize the entirety of the human scientific endeavor. Most of us don’t appreciate what that means.What a time to be alive!
Sep 2026
Imagine where AI was a year ago—Claude Opus 4.5 is still over a month away.Now imagine where we’ll be this time next year.
Beautiful. Bring back Art Deco!
Dume@gietzschean
Art Deco actually made a power company building look exactly like a power company.The Niagara Mohawk Building in Syracuse was completed in 1932 as the headquarters of Niagara Hudson Power Corporation.At night, concealed lights within the façade, illuminated glass panels, floodlighting on the central 'Spirit of Light' sculpture, reflective aluminum, polished glass and stainless steel turned the building into a glowing advertisement for the power of electricity in the age of machine.This also represented the optimism and beauty that could be incorporated into modernist structures.
Onward compute.
Patrick Collison@patrickc
Some data we recently assembled on entrepreneurship/compute in Europe: eudata.vercel.app.We hope that one of the useful roles that Stripe can play is in collecting and publishing empirical data pertaining to entrepreneurship and industry in Europe. There's growing appetite to get Europe on a better footing, and cross-sectional comparisons can often shine light on where opportunities lie. If you're interested in this kind of thing, we publish more at stripeeconomics.substack.com.
Who knew self-replicating AI was going to be so cute!
Kevin Ngo@kevin_t_ngo
Thank you @claudeai!GIF animated in Python and rendered in Blender by Claude Opus 5.5
This is such an enormous deal. @OpenAI 's new Luna is a testimate to this.Claude Opus 4.5 hasn't even been out for a year!
Epoch AI@EpochAIResearch
AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023.That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.
We're going to reach a point in the next year where software businesses are going to have to completely reinvent themselves from first principles.This will be hard, because both execs and senior ICs are going to have to admit that most of what they know isn't as valuable in this new world.Don't let ego prevent seeing reality. That mistake is more expensive than ever.
Talk about home gym goals.
Travis@travs35mm
These system prompts are getting crazy.
My entire feed the last few days is summarized by this lol. Take this as you will.
banteg@banteg
>be me
>discover effective altruism
>apparently normal charity is inefficient
>why donate to random sad thing when spreadsheet can tell you optimal sad thing
>fair enough
>buy mosquito nets
>save lives
>numbers look good
>feel powerful>couple years later
>someone asks an innocent question
>why only count people alive today
>huh
>future people matter too
>obviously
>my grandchildren shouldn't matter less just because they haven't spawned yet
>reasonable.jpg>keep following logic
>what about their grandchildren
>also yes
>what about people in 500 years
>sure
>5000 years
>why not
>500 million years
>starting to get weird but morality is morality>open calculator
>humanity could survive for an astronomically long time
>could colonize galaxy
>could have trillions upon trillions of descendants
>maybe digital people too
>maybe simulated civilizations
>maybe dyson spheres full of happy uploaded minds
>calculator starts smoking>realize currently living humans are rounding error
>8 billion people suddenly looking extremely beta
>future contains potentially 10^something people
>can't even fit beneficiaries in google sheets>new moral priority unlocked
>protect the long-term future>stop thinking in units of "people helped"
>start thinking in "fraction of cosmic endowment preserved">malaria?
>terrible
>but only kills existing humans
>AI extinction could delete the entire light cone
>nuclear war could permanently derail civilization
>bad institutions could lock in terrible values for ten million years
>someone invents wrong constitution in 2140
>quadrillions suffer
>better fund governance workshop now>friend says maybe we should improve hospitals
>explain opportunity cost
>friend says hospitals are full of actual sick people
>explain scope sensitivity
>friend stops inviting me to dinner>need to decide what to fund
>easy
>expected value>suppose project has one in a million chance of preventing extinction
>sounds tiny
>but extinction destroys 10^50 future lives
>multiply
>mother of god>$10 million project has expected value of several galaxies
>charity evaluation complete>someone asks where the one-in-a-million number came from
>expert judgement
>which expert
>us
>how calibrated
>extremely thoughtfully>reduce estimate to one in ten million to be conservative
>still beats curing cancer by 38 orders of magnitude
>epistemic robustness achieved>someone says maybe project doesn't work
>assign 20% chance
>still astronomical>maybe project makes problem worse
>assign 5% chance
>still astronomical>why 5
>because 30 felt pessimistic>publish 46-page report
>contains seventeen sensitivity analyses
>every sensitivity analysis begins after assuming intervention has positive sign>critic says you're multiplying enormous hypothetical stakes by extremely uncertain probabilities
>yes
>that's literally why it's important>critic says the uncertainty might be structural rather than numerical
>make probability smaller
>critic says no, I mean maybe your model is wrong
>make probability smaller again
>critic begins rubbing temples>discover AI safety
>perfect longtermist cause
>AI might kill everyone
>or create utopia
>or seize galaxy
>or tile universe with paperclips
>or create billions of conscious software minds
>finally a problem with numbers big enough for me>start AI safety nonprofit
>mission: prevent dangerous AI
>hire smartest people available
>smartest people immediately start building better AI to understand dangerous AI
>interesting>we must understand capabilities to understand safety
>we must scale models to study alignment
>we must race ahead so less responsible actors don't get there first
>we must deploy systems to learn how deployment can go wrong
>we must build the thing quickly because building the thing quickly is dangerous>outsider asks why the people most worried about AI apocalypse all work at AI companies
>complicated field>company releases stronger model
>very concerned
>company begins training even stronger model
>extremely concerned
>company raises $14 billion
>concern reaches unprecedented levels>need to influence government
>future is at stake
>normal democratic process too slow
>politicians don't understand exponential curves
>public doesn't understand x-risk
>experts must guide them
>who counts as expert
>people who understand x-risk
>who understands x-risk
>our friends>someone objects that this seems politically convenient
>explain we're representing future generations
>future generations unavailable for comment>develop concept of value lock-in
>terrifying possibility that one ideology controls civilization forever
>therefore extremely important that civilization adopts correct values before lock-in
>whose values
>let's circle back>begin with impartial morality
>end with small group of people deciding what quadrillions of hypothetical beings would want
>beautiful arc>meanwhile actual humans keep doing annoying things
>voting wrong
>having parochial attachments
>loving family more than strangers
>caring about local community
>getting upset when told their suffering is cosmically negligible
>evolutionary biases everywhere>explain that moral intuition cannot be trusted
>except intuition that future digital people count
>and intuition that extinction is uniquely bad
>and intuition that our probability estimates are sane
>and intuition that our institutional choices improve the future
>those intuitions survived peer review>someone donates $5k to local homeless shelter
>inefficient
>could have funded 0.0000000000003% of an AI governance researcher
>think of all the simulated people you just killed>okay maybe don't phrase it that way publicly>PR team says "future generations deserve a voice"
>much better>journalist asks what longtermism means
>say "future people matter"
>everyone agrees
>great
>journalist asks what follows from that
>well technically we should redirect enormous resources toward low-probability interventions affecting astronomical futures
>journalist raises eyebrow
>return to "future people matter">motte has entered the chat>critic: of course future people matter
>me: glad we agree
>critic: I don't agree that your institute knows how to help them
>me: why do you hate our grandchildren>eventually notice uncomfortable implication
>if future value dominates everything
>then helping people today mostly matters through effects on future
>education matters because future institutions
>health matters because future productivity
>democracy matters because future trajectory
>human beings slowly become instrumental variables in their own moral philosophy>see starving child
>feel compassion
>check spreadsheet
>child's direct welfare contribution negligible
>but perhaps childhood nutrition improves national institutional quality
>compassion restored>tell myself this is impartial altruism>one day assistant asks obvious question
>"how do you know your intervention actually improves the far future?">silence>open spreadsheet
>increase column width
>add confidence interval>assistant asks again
>"no, I mean how do you know the sign is positive?">stare into cosmic light cone>10^50 people staring back>none of them exist
>none of them can tell me
>none of them can falsify my assumptions>realize I have invented the perfect constituency>infinitely important
>completely silent
>and always represented by me
Man, people just don’t hesitate to cut off @Waymo s like they used to.Talk about a great gauge for trust in AI though!
If it lives up to the hype it seems like Jev from @typesafeai would make it easier and cheaper for something like an internal router to be built?
AI is this god-like thing that can comprehend the universe.Primary use case for it—updating Salesforce records.If AI doesn’t want to kill us now, it probably will after Dreamforce.
I'm unsure about paid credits for SaaS. If a company is using AI, they have a model provider. Maybe it makes sense to still charge a flat subscription and build connectors for those models?
This UI is trying to be too clever. Feels awkward to use.
Our AI future is bright. Hell yeah, brother.
Max Blade@_MaxBlade
Astra built a multiplayer lawn mow simulator with beers and cigars.Made live on the internet and my chat joined.this model is GOLD.
This is one of many amazing things world models can do. Most people don't realize how fast the future is coming at us!
Jake Steinerman 🌎@jasteinerman
The new Atlas world model is literally black magic. This shouldn't be possible from just 3 cameras 🤯
I've been talking about this trend for a while. It's becoming more and more clear this is the way foward.
Tobias van Schneider@vanschneider
Depending on how you look at it, it's both a blessing or a curse.The consolidation of roles (due to AI) means that Designers are Developers now. And Developers are Designers. In particular on front-end.In fact, I'd argue the top 20% of designers are now better front-end devs than 80% of all devs before.And in the opposite direction, 20% of the top developers are now BETTER Designers than 80% of the Designers before, with their new combined skillset.For some people this is a godsend gift, for others its their nightmare come true.For me personally, I am split here. It's both for me, because I've never wanted the role of a engineer, but at the same time I enjoy it, because I'm quite effective.But I need to remember myself: I get my satisfaction from being a Designer first and foremost, not engineering, regardless of how easy or difficult it is.When I dream, I dream of designing, not implementing my designs. I don't dream about fixing bugs, chasing edge cases. I dream of problems that are solved on the emotional spectrum. All of which still would need someone competent and skillful to implement it.Even with the consolidation of roles, I still believe that towards the top (both in reality and aspirationally), the specialization will hold true and keep existing.I don't know where I'm going, but we will find out. A have a hunch (:
Aug 2026
Curious what this means for Netflix, Paramount and all these other content-based businesses. A lot of cartoons like Rick and Morty is an interesting use case, because there's no real story arc.
Rehan Sheikh@rehan_shei
Minimax H3 Max has generates video faster than you can watch it so I hooked it to a twitch livestream! Now you can watch infinite interdimensional cable - link to the stream below
There are still so many people that don’t realize that you can use ChatGPT for work.
Everyone is talking about Sol, but GPT 5.6 Luna is an incredible workhorse. It's one of those Ah ha! moments about how affordable AI quality models are going to become.Plus if you need to run jobs that are not time-sensitive it's another 50% off.
Jul 2026
It's becoming harder for me to see how the frontier labs live up to expected valuations. It looks to me like telecom in the 2000s with enormous capex on infrastructure and R&D. But the applications on top are where the margins are.
If you think the AI psychosis is bad now. Wait until voice models start getting better and everyone is walking around with their Airpods in talking to their agents. My guess is we're about a year away from that being a regular thing.
After spending a few days with GPT 5.6 Sol. It's quite good, but feels like an iterative improvement. Not a step-function improvement like Fable.
Independence Day prequel lol?
PJ Ace@PJaccetturo
Everything they taught you about July 4th, 1776 is a lie.The Redcoats weren't soldiers. They were something else.We recovered the real footage using Utopai 🧵👇
Jun 2026
The most wild pivot of the year is not @Allbirds going into data centers. But @midjourney, an AI company, moving into step function better MRI scanners.
Midjourney@midjourney
A technical dive inside our new "Midjourney Scanner"
@ButlerRaines Strong agree. Em dashes represent how people talk and think—incomplete sentences, thoughts and exclamations. It's a wonderful way to communicate through writing. It's a shame agents caught on so fast.
May 2026
@davidsenra @jimmyiovine_ TBH, this applies to just about everything AI touches—not just music. There is going to be a lot of bullshit. But most of that bullshit is going to be better.
When AI eats all of the software, I think I'm just going to do landscape design instead.
To me @perplexity_ai is starting to look like an appealing @Apple acquisition. A 5-year-old M1 Macbook Pro is already incredibly capable with local models. Some thoughts—
If you like making things AI is just magic. You can just build anything on a whim. I spent the morning tweaking this exploding smiley face.
Apr 2026
ChatGPT 2.0 is wildly good at distilling visual concepts.
Hewar@hewarsaber
GPT Image 2 is insane for brandingDesigners, we're cooked
It took a decade for software to eat the world. We’re about to see AI to eat software in about year.
Claude@claudeai
We've redesigned Claude Code on desktop.You can now run multiple Claude sessions side by side from one window, with a new sidebar to manage them all.
@jayair Honestly, it was surprising to see Figma in here at all. If you were taking a test—which is not like the others—it's Figma at the end, for documentation. Smells a lot like busywork.
Mar 2026
Jan 2026
his weekend to I decided to coding agents another shot. Honestly, these things have made the jump from being a toy to something worth incorporating in my workflow. This is going to change how we build software and think about software businesses.
Nov 2017
Hilarious—The New Zealand chatbots designed to scam the scammers
Aug 2017
Jul 2017
Incredible long-form about checkers AI—How Checkers Was Solved buff.ly
Mar 2017
Feb 2017
More and more I realize AI will leave no stone unturned. Humans will set Params and ask questions. AI will answer.
The questions are often more important than the answers—Generative design software will give designers superpowers dezeen.com
Nov 2016
I'm waiting for what seems inevitable—AI-driven copypasta software for fake news.
I just don't buy the hype about chat bots. Then again, I'm a skeptic. buff.ly
Oct 2016
Amusing that the Philosophers in Hitchhiker's Guide to the Galaxy revolt when the super computers turns on—afraid it will take their jobs.
Aug 2016
If you want to see the future of messaging AI, look at reddit. The future is noise disguised as signal.
Jul 2016
Where machines could replace humans--and where they can’t (yet) buff.ly
Software is going to replace comedians some day—just not today.
May 2016
Apr 2016
How long until AI replaces logo designers? By these results a while. Impressive nonetheless.
The concept of AI racing is fascinating to me. Without a personality behind the vehicle, will it catch on?
Mar 2016
IBM wins Jeopardy, Google beats Go champion, Microsoft's AI, Tay, became a Hitler-loving sex robot telegraph.co.uk
Jan 2016
Has artificial intelligence really passed Go? buff.ly
Nov 2015
Artificial intelligence: ‘Homo sapiens will be split into a handful of gods and the rest of us’
Feb 2015
Robots influencing our sexual selection? I think this is how we end up with the Matrix. crockpotveggies.com
Aug 2014
Jul 2014
A man and his bot are authoring 10,000 Wikipedia entries a day. buff.ly
Mar 2014
More awesome contests—Develop a Race Car AI—I can't wait to have my own NASCAR team of bots.
Nov 2013
Great 45-minute documentary, Singularity or Bust, now available for free online.
A futurist just said that future AI will lead to a war between two human factions—Total Annihilation was right!
Jun 2013
Nov 2011
Photo: › MIT researchers develop self-programming AI video game › › What if programming a video-game AI... tmblr.co