Google’s LLM agent-based co-scientist looks interesting: Google builds AI ‘co-scientist’ tool to speed up research Early tests of Google’s new tool with experts from Stanford University, Imperial College London and Houston Methodist hospital found it was able to generate scientific hypotheses that showed promising results. And: Accelerating scientific breakthroughs with an AI co-scientist We introduce AI co-scientist, a multi-agent AI system built with Gemini 2.0 as a virtual scientific collaborator to help scientists generate novel hypotheses and research proposals, and to accelerate the clock speed of scientific and biomedical discoveries. ...

I exercise 5-7 days per week, for about 1-2 hours per day. Weights and cardio, interleaved. It’s good for now, I feel/look better, but really it’s for the future. After each workout I feel like I am making a deposit in my saving account (e.g. ideas like hormesis and antifragility and to a less degree post-traumatic growth). I’m building strength, excess capacity, that will help me over the decades to come, as long as I stick with it. ...

I’ve been using grok3 for a few days, since release (4-5 days now?). https://grok.com/ Chatting is fine. I still do ad hoc chat with gpt4o and deepseek. DeepSearch is great. I have not used this much, but I like it. I guess I don’t trust the open web. I’d rather specify sources. I LOVE “Think” in grok3. I’ve compared reasoning results to o1 and o3 mini and for my recent use cases, I prefer grok3. Taste I guess. ...

Tyler Cowen shared his reasons why he thinks AI take-off, really LLM dissemination through society will be slow: Why I think AI take-off is relatively slow I’ve touched on this before in AI/LLM Diminishing Returns, but this has more reasons and more detail. Lots of economic concepts I don’t grok. I asked for a summary via deepseek: The author discusses the potential economic impact of AI, emphasizing that while AI has significant capabilities, its integration into the economy will face numerous challenges. These include slow adoption in inefficient sectors like government, human bottlenecks (e.g., regulatory constraints), and the O-Ring model, where human limitations may hinder AI’s effectiveness. Historically, new technologies take time to diffuse, and GDP growth tends to remain stable around 2%, suggesting AI’s impact will be gradual. The author estimates AI might boost growth by 0.5% annually, leading to significant long-term changes but not immediate, noticeable shifts. Market prices also do not indicate rapid transformation. Overall, the author remains optimistic about AI’s potential but cautious about its near-term economic effects. ...

I’m writing a novella. Fiction. I’m sure you can guess the genre. And I’m leaning heavily on LLMs. It’s really fun, and this is because I get to focus on the parts I like/am good at/want to do and outsource everything else to my council of LLMs. This is a new type of writing. Like having “just-in-time ghostwriters”. So what is it I am doing? I think taste. I have an idea of: ...

I was thinking about the movie “The Blair Witch Project”. Specifically, about the progression from “kids recording a project” to “horror” across 3 acts. Firstly, let’s get a handle on the plot (via DS): The Blair Witch Project (1999) is a found-footage horror film that follows three student filmmakers—Heather, Josh, and Mike—who venture into the Black Hills Forest in Maryland to document the legend of the Blair Witch. The witch is said to haunt the woods, and the students aim to uncover the truth behind the myth. ...

I was thinking about “The Unauthorised Biography of Ezra Maas”. I don’t recall when I first read it, only a few years ago I think. (checks goodreads…) July 2023. So not that long ago. The story is good, but the multimodal nature of the story telling is really fun. Lots of footnotes. Letters. Other documents. It’s great. We have to piece the story together across the styles/tones/documents/etc. Truly “Ergodic Literature”. ...

I first read “House of Leaves” about a decade ago and loved it immediately. Specifically: Narrative at multiple levels. Found-document style Multimodal writing (letters, footnotes, mess) I also liked the core mystery: what was that house all about. My favorite parts were descriptions of measuring the house, exploring the stair case, etc. When Zampano is summarizing the parts of the Navidson record that focus on Navidson + family + friends trying to understand the house. ...

Normally, I’m a “systems over goals” kind of guy. Nevertheless, sometimes having a goal kicks ass. It cuts through everything with questions like: Will reading this get me closer? Will doing this get me closer? What’s one thing I can do to get closer? I’m not saying “goals over systems”. I’m still a systems guy. I’ll system my way all the way up and over the goal. Just sometimes, we need a sharp knife to cut away the cruft and a crisp goal can be that knife. ...

I was talking with LLMs about fiction that I might like to write. I think we may have settled on “metafiction”, defined via gemini: Metafiction is a form of fiction that self-consciously addresses the nature of fiction itself, often by blurring the lines between reality and fiction or by exposing the mechanisms of storytelling. My favorite stories are those where the plot is vague (so I can guess and think), the narrator/s are unreliable, there is a blurring of fiction/non-fiction, and often a mixture of formats/form factors (footnotes, letters, diaries, editor notes, etc.). ...