about us
AI Safety Delhi is a small group of engineers and researchers who think the most important technical problem of our time is badly under-attended in India.
We started in Delhi after noticing how much of this work happens in a handful of labs abroad, and how little of it happens here. We are not a lab and not a think tank. We are three people doing research, reading closely, and bringing others into the field.
If you are already working on this, or want to be, we would like to hear from you.
organisers
A
Abhishek
organiser
Hi, I'm Abhishek. I come from software — I build things, this site included — and I spend most of my time on the technical side of what we do: reading papers, trying things out in small experiments, and turning what we learn into something the next person can pick up faster. What pulled me in was noticing that the arguments for taking this seriously kept getting stronger while the number of people working on it here stayed close to zero. Happy to talk about alignment research, how to start in the field without a formal background, or anything you're building.
V
Vivek
organiser
Hi, I'm Vivek. I work on the research side: reading closely, arguing carefully, and trying to hold the technical claims and the strategic ones to the same standard. I'm less interested in whether AI will be transformative than in the specific mechanisms by which it could go badly, and which of them we actually know how to work on today. If you disagree with something on this site, I would especially like to hear from you.
R
Rajat
organiser
Hi, I'm Rajat. Much of my work here is the part that isn't research: finding people who should be in this field and aren't yet, running the reading groups and sessions, and keeping the door open for people arriving from outside machine learning. I think the shortage in India is one of attention rather than talent. If you're curious but unsure whether you're qualified, message me — that question is usually the wrong one.
Three of us, for now. If the problem pulls at you too, write to us.