Sami Jawhar

Neurotechnologist, serial entrepreneur, digital nomad, builder of things, wannabe philosopher

[LASR Labs @LASRlabs]

DEADLINE ANNOUNCED: Apply by April 26th to LASR Labs AI Safety research program. This is a 3-month paid opportunity to write a paper in a small team with expert supervision. Past work has been published at workshops/conferences. Previous alumni now at Open AI & UKAISI.

[Sami Jawhar @CyberMonkSam]

Good evals need good infra. We're hiring engineers. Come join us!

[METR @METR_Evals]

When will AI systems be able to carry out long projects independently?

In new research, we find a kind of “Moore’s Law for AI agents”: the length of tasks that AIs can do is doubling about every 7 months.

[METR @METR_Evals]

METR is running a pilot field experiment to measure how AI tools affect open source developer productivity.

If you're an open source developer who wants to make $150/hour to work on issues of your own choosing - consider expressing interest 👇

[METR @METR_Evals]

Can frontier models cost-effectively accelerate ML workloads via optimizing GPU kernels? Our take at METR: yes, and they’re improving pretty steeply – but it’s easy to miss these capabilities without good elicitation and “fair” compute spend.

[METR @METR_Evals]

How close are current AI agents to automating AI R&D? Our new ML research engineering benchmark (RE-Bench) addresses this question by directly comparing frontier models such as Claude 3.5 Sonnet and o1-preview with 50+ human experts on 7 challenging research engineering tasks.

[METR @METR_Evals]

How well can LLM agents complete diverse tasks compared to skilled humans? Our preliminary results indicate that our baseline agents based on several public models (Claude 3.5 Sonnet and GPT-4o) complete a proportion of tasks similar to what humans can do in ~30 minutes. 🧵

[Sami Jawhar @CyberMonkSam]

@elicitorg Maybe a bit too concise? I've noticed a tendency for the section summaries to mention the topics covered in those sections but not much of the views/arguments expressed by the authors on those topics.

[Sami Jawhar @CyberMonkSam]

@james_elicit Especially appreciated the nuance about optimising within the *known* landscape. Promoting the meme of safety as a "windfall" might be a surprisingly effective cognitive vaccine against risk-taking behaviour among practitioners.

[Sami Jawhar @CyberMonkSam]

@smfleming great talk at TCPW, and an exciting response to my question about the link with awareness/intrusive thoughts. Looking forward to exploring that connection further

[Sami Jawhar @CyberMonkSam]

@AILA_Community @RTI_Intl @AllofUsResearch Will demo day be open to the public?

[Bryan Johnson @bryan_johnson]

We received FDA authorization to measure ketamine’s psychedelic effect on the brain using @kernelco Flow with our partner @CybinInc

[John Griffiths @neurodidact]

Anyone looking for a broad and deep perspective on current computational neuroscience & its intersection with AI could do a lot worse than going through the entire back catalouge of @pgmid 's BrainInspired podcast https://t.co/lphViYmmS9

[Bryan Johnson @bryan_johnson]

Announcing @KernelCo Flux: the World’s Most Powerful Brain Interface

https://t.co/r0fbZfoYiP

[Sami Jawhar @CyberMonkSam]

@docqhuys @introspection @ThomasWolfers @michaelbrowni19 @mpwpaulus @amarquand Fantastic talk! Thanks @introspection !

[Sami Jawhar @CyberMonkSam]

@dileeplearning @pgmid @ProfData Loved this talk (and the rerun on @neuromatch). Have been captivated since the CHMM paper at @CogCompNeuro. Thinking now about how "HC as inference on cloned cognitive graph" fits so well with "default mode network as MDP" from https://t.co/0fJkOSdiBN

[Sami Jawhar @CyberMonkSam]

It's rough, meandering, and overlong, but at least I finally finished getting my thoughts down in a somewhat intelligible form. Something I've been thinking about for a while: what does the mind process tells us about responsibility and justice?

https://t.co/shptPJLEL6

[Sami Jawhar @CyberMonkSam]

@gili_karni Not a mailing list, but I highly recommend @pgmid 's https://t.co/km6qzGDJo3 podcast

[Sami Jawhar @CyberMonkSam]

@MaryamShanechi @neuromatch @seanescola @Pascallisch @bgalbraith Loved your outro talk, and just finished reading your recent motor/mood BMI review paper. Feeling inspired! I'm about to start my PhD in simulation neuro and would love to get involved in this kind of research. Any advice?

[Sami Jawhar @CyberMonkSam]

The latest in my series of articles on the "mind process" is a practical guide to mental training and mindfulness. It's not woo, it's just physics!

https://t.co/zJBPzUf7CV

[Sami Jawhar @CyberMonkSam]

TVP + PB + stevia = vegan low-carb PB Captain Crunch. My breakfast is now complete 🥣