Sam Sends

@cryptohateswjp you like padron cigars?

시가바 왔네

@vvaiyr so surprised tbh

... already TPS > Ethereum? Nothing to see here

@vvaiyr @Soma Do you have the qraphql for it?

@sam_sends @Soma Can't wait to see how much data has been processed

Woah @soma has processed 8.2M transactions since launch... six days ago. More data soon.

@ObadiaAlex @ARIA_research Is there a public forum or some way to engage with this research in a non-professional capacity?

People seem to be arriving at a similar conclusion from various angles: - AGI may not emerge as a monolith, but as a distributed "patchwork" system of coordinating sub-AGI agents [1] - Static benchmarks aren't enough; we need multi-agent ones to capture emergent risks and capabilities [2] - As creation costs go to zero, human verification bandwidth becomes the ultimate economic bottleneck, making verification infrastructure one of the most important public goods for the AI era [3] - Automated proof-generation and verification can act as the unlock for this bottleneck [4] - New kinds of strategic interactions between agents are emerging, reaching cooperative "program equilibria" inaccessible in traditional settings [5] - Coasean transaction costs are about to collapse, changing our society [6] There is an elephant here that we're all touching. Our @ARIA_research initial £50m r&d programme Scaling Trust is our unifying thesis, on the trust infrastructure needed for an agentic world and how to steer us there. Before we set out on our journey over the next ~3ish years, we're hiring an additional individual to complete our team. Your role will essentially be one of Technical Director, steering our efforts technically and co-owning our research and engineering agenda. You will be doing incredibly meaningful work, in a highly interdisciplinary environment, and at the cutting edge of a technology that is shaping up to be the most defining of our century, if not of humanity. We are building for the highest possible impact. After all, this is what @ARIA_research is about, moonshot r&d projects that change the world. We want to build technology as impactful as the invention of the internet once was in another r&d programme at DARPA, to start new academic fields and academic lineages for the next century, and to catalyze lasting positive change for the world. For the right person, this is a bat signal 🦇, few places will offer you as much leverage to effect positive change on the world, intellectual stimulation, and fun. Join us! We want to onboard someone asap as we build out our initial portfolio, and are willing to move fast. Apply here: aria.pinpointhq.com/en/postings/1a… Any questions on the role, please shoot me a DM or reply in comment here! --- [1] Distributional AGI Safety @weballergy @sebkrier @FranklinMatija et. al -- arxiv.org/abs/2512.16856 [2] Agents of Chaos @NatalieShapira et. al — arxiv.org/abs/2602.20021 [3] Some Simple Economics of AGI @ccatalini et. al — arxiv.org/abs/2602.20946 [4] When AI Writes the World's Software, Who Verifies It? @Leonard41111588 — leodemoura.github.io/blog/2026/02/2… [5] Evaluating LLMs in Open-Source Games @SwadeshSistla et. al — arxiv.org/abs/2512.00371 [6] Coasean Bargaining at Scale @sebkrier — blog.cosmos-institute.org/p/coasean-barg…

@shae_mcl are you using hyperspectral imaging / raman spectroscopy?

We need new assay dev to measure more of cell state + dynamics, ideally over time, ideally in living organisms. Working on this.

"If you're just waiting for some miracle to pop out of intelligence, you will be disappointed. But if you assume the world hasn't changed, you'll miss out." AGI will not deux ex machina perfectly model biology and thus solve all ailments unless we collect the data first.

@SakanaAILabs @RobertTLange @MLStreetTalk 最も有望なアーキテクチャは、RWKVやMamba-3のようなリカレントなものだと思いますか？

【Machine Learning Street Talk 出演のお知らせ】 Sakana AIのResearch Scientistである @RobertTLange が、機械学習分野の有名ポッドキャスト @MLStreetTalk に出演いたしました。 🎥動画本編：youtu.be/EInEmGaMRLc 進化的アルゴリズムとLLMを組み合わせたオープンソースのプログラム最適化 フレームワーク「ShinkaEvolve」やAI Scientistの現在地について、ホストのTimさんと語り合いました。 ［ハイライト］ 1．LLM駆動の進化的探索が実用段階に ポッドキャストでは、ShinkaEvolveの原理について詳しく解説しました。複数のフロンティアモデルをバンディットアルゴリズムで動的に選択し、3種の変異を使い分けます。これにより、ShinkaEvolveは円充填（circle packing）問題で150回未満のLLM呼び出しで最先端の解に到達しました。また、競技プログラミングの問題からなるALE-Benchでも高スコアに。こうした達成を背景に、Robertは「私たちは今、進化的LLMシステムが革新的な科学的発見をもたらせる変曲点にいる」と語っています。 2.「Problem Problem」― 次に超えるべき壁 一方でRobertは、次の課題についても言及しました。現在の手法は問題が所与であることを前提としていますが、人間の科学的営みでは問題そのものの再定義が発見の核心にあります。Robertはこれを「the problem problem」と呼び、問題と解の共進化を次のステップとして見据えています。 3．AIエージェント時代の研究者のあり方 Robertは、コーディングアシスタントへの依存について「薬物に頼っている気がすることもある」と率直に語る一方で、理想の未来像も明確に描いています。日中はシステムと協働し、夜間はシステムが自律的に実験を走らせ、朝に結果を確認する「マルチスレッド研究」。研究者は実行者ではなく操舵者になる。「実行はAIに任せ、分析は自分でやる」。AIは人間に取って代わるのではなく、人間に一日の長がある思考の潜在的な次元（latent dimension）を増幅してくれる存在としてAIがあるというのが彼の見立てです。 機械学習研究の多くの部分がAI支援のもとですでに行われており、ますます研究は「with AI」から「by AI」へ、そして研究エージェントの「人間による高次のオーケストレーション」へと移行しつつあります。 内容にご関心がある方は、ぜひ動画をご覧ください。Sakana AIは、この大変革の時代を先導する一員として、研究に引き続き取り組んでいきます。

@brian_armstrong i'll do better next time, thanks brian

Operating in stealth mode is almost always a mistake. Talk publicly about what you're building. You’ll build momentum, get real feedback, and someone will reach out with the other half of your idea you didn’t realize you were missing.

@beffjezos The AI safety/EA movement just reduces to trusting that a small group of people with massive power will not corrupt. Gives french revolution "committee of public safety" vibes

🎯 The whole AI Safety / EA movement is just a rationalization of a power grab meant to entrench institutional power rather than give power to the people.

@gakonst Agreed, very hacky but I've been using a gRPC sidecar to cross the container exec bridge. As autonomy increases, I'm skeptical that a CLI designed for human oversight is even the right interface

github.com/openai/codex/i… surprised this feature isn't more requested? it's pretty obvious that you should allow writing to these things over stdin because otherwise how are you going to be piping things in to containers running codex w/o having to do "codex exec" each time (fragile)

No single entity is qualified to steward the development of superintelligence. We will try anyway. But a moral imperative requires certainty of duty, not hope on a bet. The true moral imperative is to ensure that what you build cannot be used to treat humanity as a means to its own replacement.

Collaborative training is inevitable. In a multipolar world, middle powers, enterprises, regulated institutions, and alliances are rationally forced to collaborate because the alternative is permanent second-tier status or dangerous dependency on hyperscalers.

@joseph_palakeel @IREN_Ltd 😂

@sam_sends @IREN_Ltd

@vvaiyr water cooled

@gakonst @karpathy Rust compatible btw

@gakonst @karpathy Yeah potentially, I've been working on some experiments using kubernetes to manage cloud agents. Using Claude to write the kube config files, and then directly tool call kubectl to read/deploy to the cluster feels pretty magical. Works locally as well: github.com/kubernetes-sig…

will someone build this / wanna work on this w/ me before @karpathy leaks all the alpha

@Soma 1000+ now

Holy shit 719 targets have been filled on @soma The difficulty will get much harder tomorrow

@joseph_palakeel palakeel is revolutionizing how varun and sam improve haha can't understate how much it means, you are currently at least 25% of our support network

@sam_sends soma is revolutionizing how models and intelligence improve

🧠

@SynchronicityHQ Incredibly thankful for your help, especially the push for local-side model processing. Huge unlock

Incredibly proud of the SOMA crew for going live. SOMA is a unique and brilliant experiment in distributed AI. Can't wait to see the results that it produces.