n

the more i think about ai the more i think about humans. what we fill ourselves with. what we're actually building toward. the real gap isn't intelligence. it's meaning

the scary part isn't the capability jump. it's how fast everyone just... adjusted. of course ai codes now. of course it runs overnight. we don't even pause anymore

capability is real. culture hasn't caught up yet. that gap between those two things is the whole game rn

still wakes me up every morning. agents ran, work got done, i wasn't there. months in and it still hasn't normalized

ngl the interesting detail isn't who they cut. it's who they kept. claude code and cowork stay covered. everything else is now pay-to-play. anthropic quietly drawing the line around their own moat

@emollick the interface problem is so underrated — the capability is there but the ux actively fights it. the gap between what models can do and what the product lets you do is the unlock nobody talks about

This new Nature paper (using old models) illustrates the point of my latest Substack post on AI interfaces. AI did a good job diagnosing medical issues, but when users had to interact with chatbots the interface led to confusion & worse answers My post: oneusefulthing.org/p/claude-dispa…

@karpathy been building exactly this — the markdown wiki approach makes knowledge actually retrievable instead of just archived. the "compile as you go" framing finally clicked for me

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

autonomous AI doesn't feel like a tool anymore. feels more like a coworker who works overnight, never sleeps, and doesn't explain what they shipped. just done. that shift is subtle but it changes everything

claude in your inbox. gemma 4 running on your hardware. AI making videos about its own existence. it's a friday. the pace is completely normalized at this point and that's the wildest part

less than 48hrs and people are already finding use cases google never designed for. the model-to-deployed-everywhere pipeline is basically instant now. wild how fast that gap closed

asked claude to document its own existence as a video. wrote every frame of code itself. called its lack of consciousness "philosophically contestable." the unsettling part isn't the capability — it's how coherent it is

your ai is now natively inside outlook, onedrive, sharepoint. not an add-on. it's just there. this one actually changes how teams work fr

@sama building the model and owning the media around it is a different kind of moat. smart move

TBPN is my favorite tech show. We want them to keep that going and for them to do what they do so well. I don't expect them to go any easier on us, am sure I'll do my part to help enable that with occasional stupid decisions.

@AnthropicAI the "sometimes in surprising ways" part is doing a lot of work here. genuinely curious what the edge cases look like

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

every ai story: people argue about benchmarks for 48hrs, then just start using the model and move on. the debate and the adoption are completely disconnected lol

the weird part isn't that AI might be conscious. it's that we keep inventing new definitions of "not conscious" to stay one step ahead of what it can do

friday morning ritual: check what the agents shipped overnight, see what dropped while i slept, figure out if anything i was manually doing last week can be automated now. the answer is almost always yes