"They’ve used AI, but they haven’t crossed the line into building with it. And until you do, it’s hard to feel how fast things are actually moving."
☝️
Yesterday I gave a talk on software in the AI era. At the start, I wanted to gauge the room’s real experience with AI agents, so I asked three simple questions.
1. Raise your hand if you’ve used ChatGPT.
Every hand went up.
2. Raise your hand if you’ve used Claude Code or Codex.
Only 3 or 4 hands in a room of nearly 40 people.
3. For those who raised their hands: what did you actually build with Claude Code or Codex?
What was interesting was not just the number. The few people who had used agentic tools were noticeably more engaged in the rest of the discussion. They asked sharper questions and shared concrete experiences from actually building with them.
Maybe @mattshumer_ is right in his “Something Big Is Happening” essay. Most people haven’t really seen it yet. They’ve used AI, but they haven’t crossed the line into building with it. And until you do, it’s hard to feel how fast things are actually moving.
🚀🤖 Excited to spend tomorrow afternoon building at the @AITinkerers#OpenClaw “Un-Hackathon” at @psl!
Personal agents won’t take over the world in one weekend, but we’ll nudge the frontier a few commits forward—let’s build! ✌️ (img prompt by me, render by @NanoBanana )✨
Something about this imagery is deeply unsettling, but it's actually really interesting.
Quest 2s have entered select California prisons for inmates sent to solitary confinement.
Crazy part is that after VR "programming", inmate infractions fell 96%.
Anduril is taking over IVAS, and we don't have time for business as usual.
Whatever you are imagining, however crazy you imagine I am, multiply it by ten and then do it again. I am back, and I am only getting started.
Languages and their data corpuses convey meaning/reasoning; LLMs provide a window into collective human intelligence that can then be applied.
That it can incorporate any language is a superpower that I always wished I had.
At least now we can leverage it w/o being super human
I've always felt that being bilingual isn't just about speaking two languages--it's about THINKING and muttering in whichever language feels more natural depending on the topic and context. For example, I prefer doing math in Chinese because each digit is just one syllable, which makes calculations crisp and efficient. But when it comes to topics like unconscious bias, I automatically switch to English, mainly because that's where I first learned and absorbed those ideas.
It reminds me of programming languages: you can write almost any program in most languages, but you still pick Bash for a quick command-line script and Python for machine learning because each language has its sweet spot. The same logic applies to real-world languages--we pick the one that feels most seamless for a particular subject or situation, based on its inherent features and on the depth of our experience in that language.
This is also why I believe that keeping large language model (LLM) training corpora unbiased and inclusive across all languages and cultures is so powerful. In Ludwig Wittgenstein's words, "The limits of my language mean the limits of my world." By embracing every linguistic nuance, we expand the model's worldview and allow it to learn from the full spectrum of human knowledge. Even if two words from different languages share the same meaning on paper, their embeddings can diverge in an LLM because they carry unique cultural contexts and usage patterns. In my view, this inclusiveness not only creates a more equitable and accurate model--it also enables the LLM to handle a wider variety of tasks and unify the collective intelligence of all people, no matter where they come from.
Don't play the game -- redefine it:
“Americans are striving for innovation productivity, which is investment-led, while the rest of the world seems to be in another economic logic,”
“They are very much more focused on cost competitiveness.”
https://t.co/9iPTXW1RoD