The events of the last 6 months in technology are arguable amongst the most important in human history
The tools now increasingly exist for recursive self improvement of models & agents
We are likely in very early lift off & exponential
Largely unnoticed outside of tech
@nummanali Cowork addresses the "AI agent for business" usecase, running in an isolated VM for maximum safety for non-developers. Codex does not seem to address this market
@notgoing2argue@MikeTruesdale5@Andercot The H is a dual intent visa, meaning that if enter US on H and say at the border that you intend to immigrate and stay permanently in the US, they can't stop you. While H is non-immigrant, you are legally allowed to apply to become an immigrant without going back
@MikeTruesdale5@Andercot It depends on the work visa. Some work visas, such as H and L, are legally stepping stones to green card. The US law says that you can apply to and obtain such visas with the intent to immigrate permanently. Other work visas are purely temporary (for instance J).
@sporadica The correct read is that people hired in non-technical roles now have "technical" expectations. If you ship code, you are de facto technical.
@eshanbuilds@DavidSacks The economics have changed but you can't blame one model for it, as all models will get there in 6 months. You can no longer assume adversaries don't have access to such models.
@wesley_8@bindureddy Since January every lab is focusing primarily on coding, including OpenAI. Expect the next version of Gemini to be targeted primarily at coding.
@landforce@chris_grant The sandbox is also a big difference. Claude CoWork runs in its own virtual machine and can only damage the folder you gave it access to. Claude Code can in theory hack your entire computer.
@chamath The ideal for every engineer is to build 0 to 1 products... those in which the code can be written from scratch without prior baggage and constraints. In the corporate world, it is extremely rare for engineers to get 0 to 1 projects.
@dysmemic@lateinteraction@davidbessis LLMs lack many of the components that define human intelligence. Even so, with the right harness they can be superhuman in getting things done with high precision and speed
@erikrcarlson@karpathy I tried to warn the leaders at my son’s school that every teacher should be actively using agentic AI now, and my warning got 0 traction. To a group that is setting the AI strategy of the school
Hello, Moon. It’s great to be back.
Here’s a taste of what the Artemis II astronauts photographed during their flight around the Moon. Check out more photos from the mission: https://t.co/rzM1P0QbOl
@jack Surprisingly tone deaf for an org leader. The reason orgs have a low manager-to-employee ratio has nothing to do with bandwidth. People need connection and support to operate at max capacity.
@SullyOmarr Yes, as a backup. When the tools don't work, and also browser access fails, as a last resort Claude can get the thing done using your own browser via computer use. Better than not getting the thing done.
@DannyMagazu OpenClaw has a much superior user interface. Hardcore engineers will find Claude Code more useful because they can configure it. It's like comparing Google Docs with vim.
@andrewchen@MisterMarket0 The value of spreadsheets is in the mindshare, and that is really hard to change. Standardization on a tool that everybody knows how to use is irreplaceable