I've been reading AI As Normal Technology by Arvind Narayanan and Sayash Kapoor: https://knightcolumbia.org/content/ai-as-normal-technology. You may know them as the people behind the AI Snake Oil blog.

I wanted to open up a discussion about their concept-cutting of AI as "normal" technology, because I think it's really interesting, but also gets a lot of stuff wrong.

Oh, if you read some of Plato's dialogues it seems very untrue...Plato was really into strawmanning his opponents' arguments unfortunately :)

Anyway. To try and answer your (very thoughtful) question:

• Get people from different disciplines together in the same physical space on a regular basis. Maybe you put the software engineers next to the literary critics and get them to have lunch together regularly, or something. People are easier to relate to up close.
• Get people to work together on big interdisciplinary problems such as satellite imagery for conservation, reducing light pollution, heck even recreating space missions in video games. The key is to get a problem that needs at least 2 different kinds of expertise to address it. Then people with expertise in different areas will come together and discuss stuff with each other.
• Related: more flexibility in funding! When I was an interdisciplinary researcher in science and technology studies, it was an absolute headache trying to get funding to help me finish my PhD. Because I worked on the history of science, I was too "sciencey" for a lot of humanities funding but my work barely registered to science funders. In my experience, interdisciplinary research tends to fall through the cracks and as a consequence it's more difficult to find funding. I would like to see more explicit commitments to funding interdisciplinary research.

ChatGPT-4 and Claude. I use them for help drafting administrative paperwork like grants and they massively help with pre-existing executive functioning issues.

I find them most helpful for giving me ideas or a template I can work with, rather than just using their vanilla output.

This is a great writeup, thank you! I did my research on ethnographic observations of more "traditional" science institutions. There's still so much we don't know about science as practice, and I'm glad that this new breed of scientific institution exists to help us know more about it.

Question for you - if you were going to extend your research, what kinds of research questions would you be looking at exploring?

[crossposted from my blog; some reflections on developing different problem-solving tools]

When all you have is a hammer, everything sure does start to look like a nail. This is not a good thing.

I've spent a lot of my life variously
1) Falling in love with physics and physics fundamentalism (the idea that physics is the "building block" of our reality)
2) Training to "think like a physicist"
3) Getting sidetracked by how "thinking like a physicist" interacts with how real people actually do physics in practice
4) Learning a bunch of different skills to tackle interdisciplinary research questions
5) Using those skills to learn more about how different people approach different problems

While doing this, I've come to think that identity formation - especially identity formation as an academic - is about learning how to identify different phenomena in the world as nails (problems with specific characteristics) and how to apply hammers (disciplinary techniques) to those nails.

As long as you're just using your hammer on a thing that you're pretty sure is a nail, this works well. Physics-shaped hammers are great for physics-shaped nails; sociology-shaped hammers are great for sociology-shaped nails; history-shaped hammers are great for history-shaped nails.

The problem with this system is that experts only have hammers in their toolboxes, and not everything in the world is a nail. The desire to make everything into one kind of nail, where one kind of hammer can be applied to every problem, leads to physics envy, to junk science, to junk policy, to real harm. The desire to make everything into one kind of nail also makes it harder for us to tackle interdisciplinary problems - ones where lots of different kinds of expertise are required. If we can't see and understand every dimension of a problem, we haven't a hope in hell of solving it.

The biggest problems in the world today - climate breakdown,  pandemic prevention, public health - are wicked problems, ones that can't be easily reduced into nail-shaped problems for hammer-shaped solutions. They require lots of different problem-solving approaches for lots of differently-shaped problems.

The question, then, becomes how we stop trying to make everything into a nail - and that requires rethinking what it means to be able to problem-solve in the first place.

As I touched on briefly before, training to "think like an expert" is a set of processes that consist of looking at phenomena, figuring out how to approach those phenomena, and figuring out how to arrange these phenomena into something nail-shaped. Becoming recognised as an expert is about being able to demonstrate to others that you can take a phenomenon, form it into a nail shape, and apply your disciplinary hammer to it. If you are very lucky or work in a unique field, you might be trained to see phenomena as having more than one "shape" and more than one useful toolset, but this is still less common than it should be.

What's more, as different disciplines mature, we start finding out new things and working in the fields get ever more complicated. This means that when it once might have been possible to apply the disciplinary hammer to a wider set of problems, the maturation of a field means that the set of problems where you can usefully apply the disciplinary hammer becomes smaller and smaller. We have an ever-larger set of problems and most of our tools are wildly unsuited for solving them.

How, then, do we move forward? We certainly can't close the proverbial stable doors after the horse has bolted; the only way to stop fields from maturing is to stop people from working on them entirely. Good luck with that.

Instead, what we should do is put people working in wildly different fields in the same social spaces. We should encourage them to talk to each other - not just about work, but about life and love and random trivia - so that we can understand each others' perspectives. We should prioritise interdisciplinary and transdisciplinary work, as some funding bodies are already doing. And we should really listen to each other and be explicit in how we try to approach problems. Maybe then we can stop seeing everything as a nail, because we'll have more than just a hammer in our collective toolbox.

I do feel bad about killing mosquitoes by squishing them. On the other hand, I wear mosquito repellent, try not to leave standing water around, and I'm quite happy to keep geckos in the hope that they kill and eat the mosquitoes. Clearly, my revealed preference here is that I don't like killing mosquitoes directly but I'm happy for them to die indirectly.

I find killing mosquitoes - and generally causing suffering to mosquitoes - to be upsetting, but I find malaria and dengue even more upsetting. Beyond that, I just don't like horrible itchy mosquito bites.

Hi all! My name's Osnat. I have been interested on and off in effective altruism since my teens, but only just signed up for the forum.

I've been interested in science, ethics, and science as an ethical pursuit since I was very little. I have a background in physics and I trained as a space historian (it's a real job!)

Since finishing my PhD last year I've decided that

a) I can put my skills to better use outside academia than inside it

b) since I know quite a lot about how science is done in practice, thanks to my science-y background and science-adjacent research, I want to work on tools to make science more effective

c) science (and indeed most areas of life) would benefit greatly from more forecasting

d) I would do a decent amount of good by trying to make forecasting more accessible to policymakers and scientists. Not a tremendous amount of good, but more than I was doing before. And I think one has to start somewhere.

So, now I am working on testing free, open-source prediction market software and on engaging people with forecasting.

I write about stuff and things at osnatkatz.com.

Dang, I wish I'd read this before I put in my Manifund application - and before I chucked in some money towards other Manifund proposals. This is very insightful. I hope to participate in more Manifund funding efforts, and hope to take your advice to heart going forward.