Branch 03 of 04
Incessantly Learning
I’m forming my worldview of the future.
What I’m learning right now
The near future of AI
What the next one or two years could look like, and how progress in agents, alignment, and automated research might shape it.
Vision-language models
How different training paradigms change what models pay attention to, what they ignore, and which shortcuts they learn to rely on.
Robotics
How close robots are to automating different kinds of physical work, and which parts of the world they will change first.
Brain-computer interfaces
How brain activity can be understood and eventually translated into thoughts, intentions, and words.
Startups and products
How companies get started, find the right problem, and learn whether people actually want what they’re building.
Languages and cultures
Learning French, understanding what makes different places unique, and slowly preparing to visit as many of them as I can.
Channels I learn from
Questions I keep coming back to
Questions I keep coming back to
- What comes after LLMs, and what are they still fundamentally unable to do?
- Where are we on the path to RSI, and how far away might it be?
- How far away are robots from automating useful work in the physical world?
- Will AI change society like the internet, the Industrial Revolution, or something even larger?
- If AI becomes better than humans mentally and physically, what will still distinguish us?
- Will personal agents become the main way people interact with the internet?
- What happens to software startups when anyone can ask an agent to build exactly what they need?
- How do we make AI genuinely useful and approachable for the majority of the world?
- What should we build today for the world AI is creating?
If you’re interested in talking about any of these, reach out.