(Written with AI assistance and published with minimal editing just to get the thing public rather than procrastinating!)
Hello!
Recently I made a v1 of EA AI Uplift → a site to find people helping EA organisations use AI (& independent contractors), & assess your own AI use & gaps in your practice (and your org's practice).
The site is a v1 → part of the reason for sharing it now is to find the people and useful work that I'm missing.
(By “AI uplift”, I mean helping people use AI effectively in their existing work. That might mean making a recurring admin task less painful, helping a researcher build a useful workflow, or getting an organisation's knowledge into a form that its AI tools can actually use.)
Why I think this is worth more attention
I think there's a pretty strong case that:
- More AI adoption at EA orgs could be beneficial in various ways
- There are hardly any independent AI uplift people right now and we could use way more
- We're early enough that e.g. an in-house AI uplift role is still rare
There are several jobs involved: finding a task worth improving, making something that works, checking it, and helping colleagues use it. Buying everyone a subscription doesn't do all of that. Nor does having one enthusiastic person who occasionally shares a good prompt.
For some organisations, this might justify a dedicated role. For others, it could mean protected time for an existing team member, or a contractor helping with one bounded piece of work. I don't think we know enough yet about which arrangements work best.
The case I'd like to test is fairly straightforward: can we help people do more of their useful work, at sufficient quality, after accounting for the time spent setting things up and checking the output?
That seems worth testing on real tasks. It also seems worth sharing the failures, including the tasks where AI added more work than it removed.
What's on the site
- People doing this work. The directory currently lists 27 people, including people working in-house and independently. You can browse by cause area and organisation. Availability varies, and a listing isn't an endorsement. Some profiles are clearly labelled public-source drafts that the person hasn't reviewed.
- Personal and organisation assessments. Ten questions to help identify gaps and a useful next step. These are early working models, not validated measures of AI competence. Answers stay in your browser unless you choose to save or share them.
- Practical guides. These include a first useful task, shared organisational context, connecting tools, checking outputs, and a practitioner pathway that ends with handing work over to someone else.
- Routes to get help. Contact a practitioner directly or propose a public request. There aren't any live public requests yet.
- A place for case studies. No case studies are published yet. I'd like these to describe what changed, the evidence for that, and what didn't work.
More people learning to do this
Another reason I made the site is to give people a route into this work.
If you're interested, I'd start with one real, bounded task. Improve it, check whether the result is useful, and see whether another person can use what you made without you sitting beside them. The practitioner pathway gives a more structured version of that.
I don't know how large the paid market is. The directory isn't a census, and it doesn't establish a shortage of practitioners. But making people's work easier to find, and their methods easier to learn, seems like a useful place to start.
What would help
If you're already doing this work, add yourself or suggest a correction. If you've tried it inside an organisation, I'd particularly like to hear about one workflow that colleagues kept using, or one that didn't survive the initial experiment.
And if you think this is the wrong problem to focus on, I'd be interested in the objection.
Disclosure: I maintain the site and offer paid operations and AI-adoption work myself. I'm listed alongside the other practitioners. The directory isn't a team or an official EA body. More about me → https://www.alexanderlarge.com/legible