One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.

think about this a lot these days

I'm thrilled to announce @nova_ai's $40M funding across a Series A led by @chemistry and a Seed led by @Accel, with participation from @conviction and SAP's venture arm SAP.iO.

🚨BREAKING It looks like ICE just shot and killed someone else in Minneapolis.

"LLMs writing about the experience of being an LLM" is a moving and fresh genre of writing. This is my favorite example so far.

congrats to the very special team @cognition (@ScottWu46 @stevenkplus1 @walden_yan) on their newest fundraise -- and their big deployments turns out competitive programming (and density of intelligence) does translate well to company-building

We are opening up a new role at Quora: a single engineer who will use AI to automate manual work across the company and increase employee productivity. I will work closely with this person.

I wish more people were taking seriously the possibility that @ezraklein and the leadership of the AI labs are raising: that AGI is a real possibility in the near future. You don't have to buy it yourself, but leaders & policymakers need to consider the possibility it is true.

Interesting! Based on “The @AnthropicAI Economic Index” (millions of Claude conversations) AI usage is currently concentrated in software development and technical writing. TL;DR: 💻 Computer & Mathematical jobs dominate AI usage (37.2% of queries) despite being only 3.4% of workforce 📝 Writing/editing tasks are second most common (10.3% of queries) 🤝 AI augments human work (57%) more than automates it (43%) 💼 36% of jobs use AI for at least 25% of tasks, only 4% use it for 75%+ of tasks 💰 Mid-to-high wage jobs show highest AI adoption, both lowest and highest wage jobs show minimal usage 🔄 Tasks are being enhanced rather than jobs being fully automated 📊 Analysis based on real usage data from millions of Claude ai conversations 🎯 Usage concentrated in specific tasks rather than entire occupations

The 80/20 of AI Agents is basically: - know xml - know what a webhook is - know how to use an api - be good at prompting Anything beyond that is just model capability + creativity.

Personal highlights from Claude's snarky AI comedy set.