Notes 2021, episode 4
Open source, new people, and donkeys
Last month started with one of my favourite things in the world: releasing open source code. The Data Lens project is something that had been agreed and initiated before I started, but that resonated a lot with my core passions: data, search, applied NLP, and replicable platforms. It was the pilot project for my team and programme, and it tested a lot of what we needed to test: the ability to act rapidly, using applied AI on problems that the people working in the wider NHS system presented us with, releasing open source not just as a tick-box exercise but as a way to commit to openness and engagement through the building of AI toolkits.
This release was followed by two announcements in one: the publication of the list of projects selected through our problem-sourcing initiative (I don’t like to call it a competition), and the kick-off of the next round. This part of the Skunkworks programme works mainly through matching suppliers to the problem through the ACE framework. We ask for a strong commitment to release open source code for any project that creates a proof-of-concept, as we’ve done for Data Lens, and I’m really looking forward to selecting the suppliers that we will unleash on these problems, which as you can see combine clinical practice and admin.
Also, you know me, I will not shy away from addressing this tweet:
You see, Charlie raises some great points, and I fundamentally agree with him: recruitment is on record as a potentially troubling use of AI. But the question that was posed to us was to discover if this approach can work at all safely and ethically, or if we can learn — by working together — that there isn’t a way to do that. This is one of the key elements of my view of what a Skunkworks team does. We’re at a stage of AI adoption that requires experimentation, hand-holding, scrutiny, and debate. Our overall mission is to understand how to deploy AI safely and ethically — which means also learning and being open about difficult situations. There are plenty of suppliers who offer AI as a recruitment tool. We want organisations in the NHS to acquire knowledge about the problem, the issues, and potential solutions. By working on this project, we can offer all that in a controlled environment, getting the oversight of my colleagues in the AI Ethics team.
The team is growing, too, which is very exciting :) I have 3 new joiners in the next few months. My first Senior Data Scientist started a few weeks back, which has given us an opportunity of returning to some of the problems that we didn’t select. This is to start working on the other big part of the Skunkworks strategy: providing hands-on AI capability to organisations in the NHS galaxy that want to learn more about it, understand which problems are good candidates to apply AI, and work with us on discovering if that’s a good idea or not.
People-wise, it’s been a solid month of fantastic conversations. I’ve been having some great chats with Andy and ian about projects we’ve run in “lil’ old Kettering” and potential developments of them. It’s great to see so much talent and progressive approaches to good healthcare everywhere in the country and not just in London. I met with Niahm for an outdoor coffee and chat, which was lovely after having worked together on a project for a few months. A good chat with Tara and Alice on digital initiatives for mental health left my brain buzzing with excitement. I also joined Traveline’s board as an observer, and I’m really enjoying keeping in touch with the transport world, especially the data-savvy bus world, and Julie’s mission of making sure bus data is reliable.
I spent a few days in the New Forest with a few friends (yeah, all in separate B&Bs to comply with regulations, but catching up for nice walks or in freezing beer gardens in the evening). It’s been a much needed holiday, and I got rather acquainted with a few locals (or, rather, they decided they wanted to get acquainted with me by pestering me at every corner).
Allotment work resumed, too, and the digging up of the new plot is in full swing. This year — our first in this site — we’ll probably spend a lot of time clearing beds, so the planting will be minor, but I’m already loving the plot and its lovely views onto Alexandra Palace (there must be a joke in there about losing the plot?).
In newsletter news, I had been flirting with the idea of external contributions for a while, but then it occurred to me that what I really wanted was to hear from other data geeks how they work with data, in a lightweight manner. So I’ve started a new section with short 6-question interviews, copying from what Dense Discovery does very well. The first interview will be in next Tuesday’s issue, with six questions to Sophie Warnes. You can always subscribe here.
See you next month!
