Alex Tan
I'm a Stanford undergraduate from Portland, Oregon. I got into technology by trying to get around its limits. These days most of my attention goes to AI, and specifically to the part after you hand it some work: seeing what it did, checking whether it was right, and stepping in when it wasn't.
Work
- Scrollable. A browser exercise for practicing agent supervision. You predict what an agent's code will do, watch it work through made-up requests, then check the outcome against what actually changed.
- AltOS. A personal planning system that connects my calendar, tasks, and longer-term goals. The AI can propose changes, but nothing touches my calendar until I approve it.
- AI reliability experiments. Small evaluations of two open-source Stanford research systems. In one synthetic test, a speech recognizer that made fewer transcription errors led the assistant to carry out the wrong action more often.
There's also Delegation Lens, a small interactive sketch of reviewing an AI's proposed change before approving it.
Interests
The question behind most of this, and one I don't have an answer to, is whether working alongside AI makes people sharper or quietly does their thinking for them. It's part engineering and part philosophy, so I end up spending time on evaluation and interface design, and on older questions about minds and consciousness.
Away from a screen I play basketball, sketch, and have a habit of turning ordinary conversations into philosophical rabbit holes.
Contact
Email is the best way to reach me: alex@alextanm.com. I'm happy to hear from anyone thinking about similar things.