Hanging out my shingle

1 The most urgent thing

I need to scope what I mean by impact. For these purposes, it’s AI safety. I think we’re mishandling every aspect of the transition to an AI-driven society. Many parts of this transition will probably go awry unless we invest in proper mitigations, and right now we’re not doing well.

My risk model here is broad.

• There’s a risk of perpetuating or worsening inequality across society.
• We have a real chance of losing the delicate human understanding that underpins liberal society and democracy.
• There’s also the danger of total epistemic collapse: we could lose the ability to understand the world and make good decisions, drowning in a sea of AI slop.
• Moreover, we face psychological risks of becoming alienated as machine intelligences infiltrate every aspect of our lives.
• We risk being shot by autonomous, weaponized drones or hit by AI-engineered bioweapons as geopolitical tensions rise.
• There’s the looming threat of autonomous intelligent algorithms spiralling out of control and causing untold harm to humanity for their own reasons.

We need to address all these issues, and quickly. The time to work on improving this situation is not after a disaster strikes, but before it does.

There are many other massive problems in society, but this is the one that will hit us first, because software moves so damn fast. We need to get it solved in order to work on the others.

I think it is very much worth working on AI safety now, because there is also a massive potential upside to getting it right. If we can get trustworthy, reliable versions of these tools in the hands of the people who need them, they could help us solve all the other problems we face, e.g. climate change, democracy, health, global poverty…

2 What I am up for

I could use ideas, collaborations, roles, co-founders, mentorship. New contacts. Old contacts who didn’t know I was looking. I want help to work out what I should do next to make a difference on this problem. Maybe we should work together?

I’m open for business, up for ideas, and ready to start things. I’m ready to join your project. I’m ready to found a new project. I’m looking for collaborators, co-founders, and employers. Solve an important scientific problem? Persuade an important institution to change its ways? Build a new institution?

The things I pursue should be impactful. Everything’s on the table right now. I’d work in your company, work for the government, or found a new venture. The key point here is that I’m prepared to trade prestige, esteem, and seniority for a chance to make tangible improvements.

Does that speak to you? Get in touch.

If you’re about to suggest some random non-impactful role, thank you kindly! But I don’t need help with such roles; I can find ’em myself.

3 What I’m useful for

Vanity deep research is the new vanity search! Here is what the algorithms say about me; tl;dr I’m some kind of futuristic Swiss army knife of AI research and strategy.

• Gemini summary of me
• ChatGPT summary of me

My key value proposition is that I’m good at several things you rarely find in the same person, which makes me excellent. Bluntly, my combination of IQ and EQ and domain expertise is rare. But let’s get into detail.

AI safety is something I am very qualified to help with, as a machine learning expert with a broad view of the landscape and a lot of experience at the coalface. I’m a voraciously curious, smart guy who’s good at hard intellectual work, especially mathematical work. I’m highly motivated by difficult problems and interesting intellectual challenges. I’m driven by working with people towards shared goals. Projects that bring that enthusiasm to bear will go very well. My enthusiasm is a sandblaster which can wear down granite problems.

As my current² role (research scientist in a major ICT research lab) hopefully indicates, I’m a skilled artificial intelligence and machine learning researcher with a considerable ability to produce cutting edge AI methods. I understand the mathematics of deep learning, linear algebra, probability theory, and physics modelling, and I’m one of those people who invent the algorithms that make the AI world progress. I publish papers in top-tier machine learning conferences, pushing forward the state of the art in these areas. In the world of machine learning research, my particular skill is breadth; people come to me when they need the right tool for a weird job, because I know many, many tools. If you want to know how to get 3% more performance out of your LLM training, you wouldn’t come to me. If you want to know how to replace your entire training pipeline with something radically better, I’m your guy.

I’m a qualified statistician and data scientist, with qualifications from a range of reputable universities. I have a deep understanding of, and experience in, statistics, probability, and causal inference. I can design experiments, analyse data, and draw valid conclusions. I’m trained in econometrics, ecology, geospatial data analysis, and epidemiology.

People skills: yes. I am intensely pro-social and collaborative, and I can get many people to argue that when I am on a team the team works better. If that doesn’t seem valuable to you, then we have very different models of where high performing projects come from. I can run projects involving people. I’ve managed lots of students, supervised a PhD, and run a number of conferences, workshops, and events. I get excellent results from teams I lead, and excellent exit interviews from people I manage. I’ve run projects in Southeast Asia, Switzerland, and Australia across multiple languages and multiple cultures. I’ve even organized bands, which is like organizing a conference but musicians are more tempestuous than baseline humans, so it counts double.

My conflict style and communication hygiene are strong. I can bring people together, resolve disputes, and get things done. I’m extremely good at getting to know people, socializing with them, and building bridges across cultural divides within and across nations, and across political positions. I’ve worked with people who have ADHD and autism; I’m comfortable working with challenging personalities.

I can code, of course. I’ve written reams of code, but I’m not going to pitch that here, because software engineering is a real and deep field, and you should hire software engineers for engineering software. Specifically, if you want some software engineered, you’d be better off hiring someone who’s spent 10 years honing their coding skills rather than 10 years honing their mathematical skills.

I know how to communicate effectively between different disciplines—between scientists and the public as well as among scientists of various specializations. I’ve been a scientist in different fields myself. I have theory of mind about how different disciplines think, and I can translate between them.

I can write. In fact, I write constantly; I’ve penned over a million words on this blog and more in various other places, not just academic journals. I can write for different audiences, from technical papers to popular articles.

I’m not afraid of being in the spotlight. If you need a show pony, I’m that guy.

4 Constraints

5 Contact

Hit me up on my contact form, or email me at dan@thisdomain.

You can also check out my funding proposals so far.