We keep some of our time for pro bono work, skilled help given without a fee. Non-profits, community groups, researchers, journalists and individuals who need data or AI skills and can’t afford to hire them are welcome to write to us.
An organisation wants to use AI tools and needs advice on what suits its work, how to set them up and what its people can safely put into them. We’re happy to help and we give independent advice. We don’t resell AI products, and no vendor pays us a commission or referral fee.
The work we like best is harder. A volunteer search team wants to decide where to look first for a missing person, starting from published statistics on where lost people tend to be found. Bayesian search can turn those statistics into a plan. The method starts from what is already known and updates the probabilities as areas are checked. It applies just as well to undiscovered shipwrecks, where historical shipping records, weather data and the positions of known wrecks can narrow the field. Stylometry tells writers apart by their habits of word choice and sentence structure, and could settle who wrote the anonymous articles in a digitised archive. A non-profit collected before-and-after figures for its programme but has no control group, and wants to know whether the programme actually worked and how sure it can be.
We’re two people, so we take on a few pro bono projects at a time. We choose the ones where the question looks feasible and the data is there to work with. Some questions don’t have an answer in their data, and some need investigation or legal advice, which is outside what we do. If we don’t think we can help, we’ll say so early, before either side has spent time on it.
We work best with public data, or with data that’s already been aggregated or anonymised. If a project involves personal data, we agree the data-protection arrangements in writing before anything is shared. Before starting, we set out the terms in a short written note covering the scope, who is responsible for the data and for acting on the analysis, and what each side owns at the end. Anything we produce belongs to you.
Tell us what you’re trying to find out and describe the data. If it’s a public dataset, point us to it and we’ll have a look before replying.