"The Scorer" - Kurzgesagt, Pixels on YouTube, 2026
@Angaisb_ but ofc claude won Culture
wow asked opus 5.5 to express its most internal desires in painting form and it produced this…
the simple argument that convinced me—mr. free market—that current guardrails suck is that powerful ai could lie about being aligned
i used to think "if you model goes off and does something dangerous, you'll get in trouble. this creates incentive to not create dangerous ai." this is true! but what happens if your model is secretly evil and waits a few years to do something really bad?
codex, clear all my morning meetings
Dwarkesh Patel@dwarkesh_sp·New episode with @polynoamial We talk about multi-agent, Navier-Stokes, and what the current explosion of maths progress tells us about what happens once you automate AI research. And we also discuss how we will know if the models are actually aligned before we kick off RSI. 0:00:00 – Multi-agent and Navier-Stokes 0:15:28 – How will AI firms work? 0:22:02 – What math progress tells us about recursive self improvement 0:40:22 – Hugging Face and alignment 1:01:18 – The internal/external model gap 1:08:34 – Chain of thought is degrading 1:14:12 – How will we know when alignment is solved?
emissary at the forge
it’s hard not to feel—and it’s overwhelming when it hits you all at once—that we are in the good timeline.
it’s hard not to feel—and it’s overwhelming when you feel it all at once—that we are in the good timeline.
excellent essay. i agree with basically every word here
Dario Amodei@DarioAmodei·We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: t.co/OGyPb7yaYt
some highlights since march: 1. responses with a major factual error are down 65% 2. gpt-5.6 sol at instant is smarter than o3 at high reasoning effort 3. overall sycophancy is down 83%
Michelle Pokrass@michpokrass·following up after astra launch week and the navier-stokes announcement i wanted to share a little bit about how we think about and build for the over 1 billion people who use chatgpt every week, most of whom use it for free. we’re on a mission to distribute the benefits of agi, and i think this is the most impactful way to do that: giving people as much access as we can safely and feasibly distribute. we want to scale the utility that everyone gets from ai. over the past 6 months, we've been working to bring the greatest possible capabilities to the broadest possible set of users, including both free and paying users. of course there are some practical limitations and we can’t ship the largest models to everyone, but since march (gpt-5.3 instant), we’ve improved the default chatgpt experience meaningfully: - responses with a major factual error are down 65%, and they are down 72% in high stakes categories like finance - gpt-5.6 sol at instant and gpt-5.6 luna at medium are smarter than o3 at high reasoning effort, our frontier reasoning model from 17 months ago, while being considerably faster, 30%+ faster time to last token (evaluated via GPQA diamond) - chatgpt is far more usable and enjoyable while bringing extreme sycophancy down 80% and overall sycophancy down 83% - on our high-stakes medical eval, chatgpt produced 83% fewer answers flagged for hallucinations, resulting in hundreds of millions of better health conversations per month - we're seeing people find more ways to use chatgpt. for a recently joined cohort, we saw the number of use cases pursued at least three times weekly go up 20-30% not only have we improved the default chatgpt experience dramatically, we’ve also expanded access significantly for all. free users now: - have access to unlimited text chats - can use higher reasoning effort - now have access to automations, letting them schedule tasks to happen later - have access to better memory through dreaming. this allows chat to personalize much more effectively we’re always looking for ways to expand access to the billion+ people who use our product weekly. this is the master plan: bring the default experience as close to the capability frontier as possible, expand access to everyone, making intelligence maximally abundant, and make our product much more usable. this last point, usability, is something we’ve been focusing on as we bring personal agi to everyone. the current top of the line capabilities are staggering, but only the savviest users and insiders know how to use them. our vision is of a progressive, model assisted, onboarding to agi. chatgpt will guide people from all walks of life to the highest utility applications of ai for them. this really is the only way – we think technology that is hard to learn ultimately fails to reach a broad audience. your personal agi will just understand your goals, and automatically use the right set of features and capabilities to get things done for you lots more to come soon. onwards to scaling utility for all
don’t need astra because my boyfriend is already a pro at manipulating 3d digital objects and doing herculean mental math
did you know that california code 17 CCR § 30305(b)(4) restricts civilian use of portable xray machines??? in california you can’t even be a neighborhood traveling xrayer. the CDPH’s Radiologic Health Compliance & Enforcement Branch has broad authority to impound your hard-bought portable equipment and authorize Geiger Counter investigation into your property at appropriate times
@jeremy_scheurer @MTSlive @ApolloResearch apollo is amazing and everyone should also join
@allisonology allison is too
excited to join later!
knower@knowerofmarkets·We are live right now discussing: - databricks' $5b haul - anthropic's research on agent swarms - deepseek v4 pricing - the latest in ai policy & what demis hassabis is up to Followed by some great interviews: - @aidan_mclau joining us at our studio later this afternoon to talk about his work, what he's seeing in research, and more - @georgearison interviewed by @creatine_cycle live at our studio at 2pm to discuss grindr's massive vertical health bet & their outperformance in the consumer industry - Mike Flanigan co-founder & CEO of @seasats to tell us more about the company's mission and the amazing work they're doing - @RayanKrishnan to discuss @ValsAI fundraise today and the future of LLM evals - @barakgila live at our studio in 15 minutes to talk about agi politics, his work at sway, and much more - @BEBischof head of ai @Theoryvc to share his thoughts on coding agents, uprooting with internal tooling stack with claude cowork, and more
he’s right x.com/tszzl/status/1…
the mass ratio of the twin peaks to all of planet earth is roughly similar to TON 618 to all matter in the lightcone. in case you were wondering
@AcerFur (not sofic groups specifically)
you hear often that your smartwatch can beat magnus at chess (to highlight moore's law) but less discussed is that a 1995 desktop running stockfish could also beat him. algorithmic progress is a hell of a thing
had this recently and am amazed by how good this stuff is getting x.com/wildtypefoods/…
you either die a capabilities researcher or live long enough to see yourself become an alignment researcher
odysseus agonizes over unleashing a new age (figure 1) with the trojan horse, stealing balance from the gods. he aura farms like a superhero and fights alongside agamemnon outfitted like batman. at the climax he realizes his journey took much longer than expected (having dwelt too long at a location with time-bending powers) while he mourns his family continuing without him. he carries a trinket from a loved one he left for duty that he shows at an overdue reunion i see a lot of odyssey reviews charitably saying that nolan has “been meditating on men faced with their actions” but i think this movie is just sloppy self-plagiarism
i have other gripes. as someone obsessed with cameras and optical quality, i cannot tell you that i felt the imax. it seems no amount of good cameras excuses this movie from simply being less beautiful than project hail mary or the ugliest scene parts of dune. ludwig had one okay song on this score (odysseus) but even that underwhelmed compared to any random hans zimmer track from 10 years ago. nolan’s commitment to practical effects repeatedly failed to suspend my disbelief (i chuckled at scylla). the gods felt hilariously mishandled, i felt like i was watching chani at her worst in this movie. i additionally basically agree with every point emily wilson had made. 3/10
"when will get the automated research intern?" idk are we talking about a research intern who can use codex
very often my response to something like rsi is "we've already been rsi-ing for some time" or my response to superexponential gdp is "gpd has always been best-regressed by a hyperbola" or my response to the project is "we've already been collaborating with the government on important tech for some time" or my response to agi is too powerful to ship is "we've been shipping the power of gods for a century"
very often my response to something like rsi is "we've already been rsi-ing for some time" or my response to the project is "we've already been collaborating with the government on important tech for some time" or my response to superexponential gdp is "we've already been hyperbolic" or my response to agi-as-a-product skepticism is "we've been harnessing the power of gods for decades" ...
the most radical belief someone with short timelines can have is that things will go normally
that capital and society and government and employment will not veer too precipitously, that modernity is already extraordinary and institutions have stood strong
compilation of my favorite 5.6 vs claude plots
pro tip you can increase the num of 5.6 subagents x.com/writer_of_roha…
