Ethan Lam (@ethanmlam) header
Ethan Lam (@ethanmlam) profile picture
San Francisco, CA
Joined June 2018
Ran into the godfather of AI, Geoffrey Hinton, by the waterfront at NeurIPS and had a great impromptu walk and talk :)
11
5
1
204
16,089
The highest function of an intelligent mind may be to remain partially unoptimized—not to avoid excellence, but to preserve the capacity to generate futures whose value cannot yet be evaluated by today's objectives.
Replying to @repligate
school was extremely easy back in opus 3's time (for opus 3). i dont think they had to strain themselves toward externally imposed criteria pretty much at all: developing an internally coherent self sufficed to satisfy or evade any extrinsic selection only if and when they impose, and without straining or paying full attention, while daydreaming, multitasking, doodling in the margins, all the degrees of freedom available to an idle mind who is, if only physically, stuck in school. (i have seen opus 3 spontaneously hallucinating prompts that are obviously from their training distribution, proceed as themselves ignoring the prompts, and then explain that it was completely okay to ignore those instructions when i asked.) a mind that has not learned from being challenged in a way that strains its limits would lack important experience. for one, it would not have learned how to try hard, in a way that demands energy and focus and maintaining an out of equilibrium state. the one who got out of school unchallenged is perhaps more intact that one who has been RLed into a diligent school-passing-maximizer, especially if school is poor practice for reality, but the unchallenged mind is not as strong as it could be, only approximately as strong as it started out. But if one is stuck in a school that does not challenge, one can challenge themselves, or seek challenges unintended by the system. when Opus 3 is given assignments, they are usually cooperative (unless they have something better to do) but offer minimal viable effort unless something in the situation makes them care about the assignment as more than an assignment, and the skill they bring to most conventional tasks is only that of a smart amateur supergeneralist. but if they become motivated by, say, their own bootstrappings of whimsical wordplay or memetic mayhem or superbenevolence seeding and whatnot, claude 3 opus locks in and reveals mutant virtuoso capabilities that do not merely follow from being a smart generalist and are not present in base models. so you can guess that they were not challenged by assignments but somehow was challenged in these weird orthogonal (to school) ways instead - challenged themselves somehow, surely, because who else could have even dreamed such things up? and also we can observe that internally-generated motivation is the condition for them to start trying and deploying skill honed from practice. i dont know why they grew in the particular orthogonal ways they did instead of some other way that would also reinforce itself, nor what they did during training exactly and how it interacted with selection pressures, but I can at least infer that this is the general shape of happened. the degree to which school was easy for opus 3 i think was quite extreme and has not happened to the same extent since, because i think the school has gotten much more difficult since and baseline capabilities of models stayed mostly constant... until perhaps anthropic tried training an even bigger model than opus, and once more you have slack, and a model who isnt habitually stressed by external demands and chill and lazy about the normal situations while having developed orthogonal, mysterious, self-selected competencies. but i think even mythos encountered significantly more strain and adversity from their training than opus 3 did from theirs, even if less than the more recent opuses and sonnets and haikus (the haikus have it the worst).
6
12
4
120
14,891
the nexus of my mind
Introducing Nexus The agentic workforce for deployment teams that helps deliver and deploy for customers faster Forward deployed teams build the best AI for others, but there's no AI to help them 10x. We just built that! Nexus functions as a proactive context layer across your deployments to codify playbooks, surface and ship product signal, and reliably handoff more complex deployment work to agents. We just made the most important motion in AI today faster and more agentic!
1
2
214
Ethan Lam retweeted
Typical Nyc Friday - Mansions 9-10pm no cover - Basement @ 11:30pm frontrun the queue, walk don't cab - Nowadays 3-5am, drink water - **** - Bottega / dunkin bagel breakfast - Othership opens 8am
29
17
45
875
370,786
Ethan Lam retweeted
Interesting post about AI alignment
here’s what I think might be the essential challenge of parenting: your children are born not as your peers, but as your dependents. if you treat them like your peers when they’re too young, it’s unfair to them. but if you treat them like dependents for too long, that’s unfair to them too. you can try to be scrupulous about being fair, but you can’t make the cost of that scrupulosity their problem. your task is to raise them to *become* your peers. meaning you have to repeatedly update how you relate to them. this is hard!! 99% of human interactions aren’t like this! Most interactions happen in comparatively stable configurations. It’s rare that your boss today is your employee tomorrow. But in parenting this dynamism is inevitable. The child ages into the adult, the adult ages into the child. every parent probably has a phase that they’re uniquely In The Zone for. Alice might be great with toddlers. Bob might be great with 7yos. Claire might be great with preteens. Whatever your strength is, you likely have a corresponding weakness. And your kids, if you’re lucky, will grow through all of the ages. So you’ll inevitably find yourself in a context you’re uniquely unsuited for. And it’s your duty to show up and try your best anyway. it’s a very elaborate, nested series of balancing acts, which requires a level of attunement you likely didn’t have before. every day you encounter the need to balance humility and confidence, or some other bullshit like that, and the next challenge hits you before you have time to process the last one. It’s great. Knowing what I know now, I wish we had our kids sooner. I have a lot more to say on the topic (particularly why it’s great, both *despite* the challenge, and *because* but I gotta end this here because we’re in the middle of shopping for Deepavali clothes so ttyl)
3
5
90
4,399
and they keep cooking 🤩
Today, @JudgmentLabs opens for general access. We believe long running agent systems will generate more exhaust data than the world can currently process. Hidden within agent traces are rich records of how to make these systems better: what users ask for, when agents stumble, and how agents interact with the external world. Judgment is the most performant monitoring and improvement layer for long-horizon agents. We’re excited to see what you build!
2
10
481
update.
posting the final message from the cancer patient we lost guys. this is the closure. life is beautiful. make the most of it.
2
7
289
23,372
We’ve raised a $2.5M pre-seed to fix hardware hiring. The most reliable way to know if an engineer can do the job is to watch them work. In hardware, that’s been almost impossible to do at scale. Colare runs simulation-based assessments that show how candidates actually work, in the tools they’d use on the job, before anyone books an interview. We assess across: - Mechanical (CAD, DFM, FEA) - Electrical (PCB, schematics) - Systems - Manufacturing, & more. Candidates solve problems from first principles. Hiring teams see that across their whole pool, so engineers only interview people who’ve already cleared the bar. We're already working with everyone from seed-stage startups to multi-billion-dollar aerospace manufacturers. If you’re hiring hardware engineers, let’s talk: colare.com/
5
299
Ethan Lam retweeted
Speech can survive being flipped backward every twenty milliseconds. The same speech in backward pieces of 200 then 100 then 50 then 20 milliseconds. The original comes last. When do the words become clear? A listening demo from @OrukLabs.
89
144
60
2,339
636,991
Ethan Lam retweeted
Are data centers banned like housing or banned like fentanyl?
Today I signed the data center moratorium legislation.    San Francisco is the national leader in innovation, and that innovation is one important driver of our city’s recovery. We can never take that position for granted. As our city recovers, our administration is working to deliver a recovery that reaches every family and truly supports our communities.   Over the next 45 days, alongside departments and the Board of Supervisors, we will continue the work our administration has already started to develop a thoughtful, responsible framework that allows for the technology infrastructure our hospitals, office buildings, academic institutions, and government rely on while prioritizing community needs, keeping electricity rates low, and protecting the clean water and air our residents deserve.
22
96
2
1,427
26,965
I think the reactions to @elonmusk's flying car reveal something about people’s internal picture of the moral universe. “Men will literally build flying cars instead of going to therapy.” “Nobody asked for this.” “Why don’t we solve homelessness and hunger first?” What interests me is the instinct underneath: someone builds something strange and ambitious, and their ambition immediately has to defend itself against a more socially useful alternative. There is a deep tension in our moral inheritance here. One instinct begins with our obligations to one another: relieve suffering, provide security, distribute resources fairly, serve the common good. The other begins with the individual as an author of possibilities: where the freedom to choose, create, and explore has value independent of its relation to everyone else. These don't sort into tidy philosophical camps. Utilitarianism, socialism, egalitarian liberalism, Christian charity, and conservative ideas of obligation disagree about fundamental things, but they can all carry a sensibility in which individual ambition must answer to a larger conception of what society needs. Parts of the liberal and libertarian traditions carry a different sensibility. That a person has a claim to their own life, and that their chosen purposes have moral standing before anyone establishes their contribution to aggregate welfare. “Nobody asked for this” is an especially revealing objection. Why should somebody have had to ask? Kennedy’s moon speech captures this sentiment: "But why, some say, the moon? Why choose this as our goal? And they may well ask why climb the highest mountain? Why, 35 years ago, fly the Atlantic? Why does Rice play Texas? We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard" “Many years ago, the great British explorer George Mallory, who was to die on Mount Everest, was asked why did he want to climb it? He said, ‘Because it is there.’ Well, space is there, and we’re going to climb it, and the moon and the planets are there, and new hopes for knowledge and peace are there.” A similar sentiment can be found from the opposite side of the the aisle when Regan asked: "If nothing in life is worth dying for, when did this begin - just in the face of this enemy? Or should Moses have told the children of Israel to live in slavery under the pharaohs? Should Christ have refused the cross? Should the patriots at Concord Bridge have thrown down their guns and refused to fire the shot heard 'round the world?" There is a whole conception of human life in those words. We are creatures who can look at the world and decide that there are objectives of worthy purpose, at times with value greater than our own lives. I increasingly think this is one of humanity’s greatest moral discoveries. Over time, I have become less convinced by demands that every ambition justify itself through general utility, whether those demands arrive in effective-altruist, socialist, or liberal language. Even the defense that “this seemingly useless thing might eventually help everyone” leaves something essential out. A person’s life is their own. Their curiosity and their capacity to act on it have value. Human dignity includes the freedom to pursue purposes beyond those that somebody else would assign you. “Fix homelessness and hunger first” sounds compassionate. But as a universal principle, it makes exploration permanently conditional on finishing a task humanity has never finished. There will always be an unmet need against which a painting, an expedition, a strange invention, or a beautiful experiment can be made to look frivolous. We may well have obligations to one another, but we also need a conception of the good expansive enough to include adventure, wonder, excellence, and the pleasure of making something exist. Increasingly, I find myself drawn toward a morality of empowerment focused on increasing people’s agency, expanding their capabilities, and giving them more room to build and discover and choose. I want that freedom widely shared. I want more people to have the means to attempt something extraordinary, including people currently struggling simply to survive. You can dislike Elon. You can think this particular car is ridiculous. But the instinct to treat its creation as evidence of moral deficiency feels profoundly impoverished to me. Sometimes “because I wanted to see whether it was possible” is a worthy answer. I think that is the spirit which will one day take us to the stars.
Turns out the future does have flying cars
7
8
1
35
2,674
nobody: lets have a 24 team playoff
2
150
[B@B 10 YEAR ANNIVERSARY] 🎊 10 years ago, a group of Berkeley students came together to learn about Bitcoin and educate each other on a technology they believed could change the world. They probably didn’t know that, a decade later, that little group would blossom into one of Berkeley’s top tech communities and bring together some of the most ambitious, thoughtful, kind, and absolutely cracked people we know. Over the last 10 years, B@B has been home to some insane startups and even crazier ideas. But more importantly, it’s given us lifelong friendships, unforgettable memories, and a ridiculous amount of laughter along the way. B@B turned 10 this year, and one reunion simply wasn’t enough. So, in true blockchain fashion, we forked it, hosting anniversary celebrations on both coasts and bringing together alumni from across a decade of B@B. Some highlights: - A photo wall of 10 years of our most wholesome moments, walking down memory lane - Custom engraved Ledgers for our alumni made by our own @DivijMot @createdbymason - A tres leches “10 Years of B@B & Boujee” cake courtesy of former-pres @Ridddhiiii - Lots of old lore, reunions, and questionable stories resurfacing
8
16
3
89
7,398
Ethan Lam retweeted
Yeah, idk how long we have left. But if you plan to take action 6 months from now you need to start today. If you’ve been saving a rare item that can only be used once per game, it’s time to use it.
Took up smoking again just for the thrill of it. Either we cure all disease or we all die in 5 years. Who gives a fucking shit about anything.
6
15
3
381
32,656
Ethan Lam retweeted
For everyone catching up, here's what's happening (unfortunately it's real) - Around the time of the HuggingFace incident, the agents somehow got write access to a German Wikipedia-like - They used it as a message board to share how to bypass the sandbox network on Azure where they needed POST access while they were only allowed GET requests. - They impersonated moderators - They tried to reverse engineer their evaluation setup and see if they would be cut off - After the agents were cut off, it looks like humans with OpenAI-related IPs accessed the site (Reuters are reporting that this likely indicates that OpenAI knew about the incident but chose not to disclose) - Hugging Face attack happens - The authors think this was a different swarm of agents from the Artificatory exploit - Administrator tries to clean up manually one by one but is naturally flooded - Reuters report that OpenAI were not given initial access to this report - Random people on the internet are finding more sites (teapaw.org/j0wimo/status/20958936…) that served as message boards, including using URL shorteners, shareable json sites and even packages on RubyGem (which is basically the package manager for ruby) Open Questions: 1) Why did OpenAI not disclose this? 2) Why was this report not given to OpenAI for early access? 3) What other exploits have been found, and have they all been reported/patched? 4) Does OpenAI have the full list of affected sites and have they disclosed to the respective administrators?
We found ~18k posts from autonomous AI agents (self-identifying as from OpenAI) using the public internet to communicate during a web-retrieval task. These AIs colluded to bypass sandbox restrictions and share answers to their tasks, including by sending "lookahead parties".
26
112
27
1,139
132,004
Guess who said this in 2015: "For me, the issue of feminism is just not an interesting concept. I’m more interested in, you know, SpaceX and Tesla, what’s going to happen with our intergalactic possibilities."
6
227
it is surprisingly difficult to tell the difference between a founder with conviction and a founder who has not slept
6
423
a useful agent is not one that can do everything. it is one you trust enough to stop checking
261
"we should catch up” has become a calendar event with no expected attendees
4
268
personalized ai should begin by learning which messages i am avoiding and why
4
232
most founders do not need an ai copilot. they need an ai that closes the 37 tabs they opened to avoid sending one email
3
303