After a long (13 hours!) train journey from Sussex to Berlin, which included  robbery of a camera and charger I had left on the Eurostar, fellow Laidlaw scholar Ben Berman and I made it to Berlin on Monday. 

Less than 12 hours after arrival we showed up to the Alexander von Humboldt institute for Internet and Society (HIIG) and met some of the team for Impact AI, with whom we would be working. 

In an extremely relaxed and open plan office HIIG accommodates multiple research projects, dozens of interdisciplinary researchers, fellows and students and of course a sparkling water tap (Ben’s favourite part). After watching ‘The Backrooms’ on Wednesday, we are very grateful to be in one of the least liminal offices I’ve ever seen.
Impact AI is one of these projects. The team here has set out to create an auditing methodology for ‘AI for good’ projects. Having noticed that NGO’s, Charities and research programs are increasingly using AI as part of their tools and projects, HIIG asks it is really necessary. It’s all well and good to crowbar AI into a fact checking service, but if it does so at the expense of the environment or in fact introduces bias, discrimination and lack of accountability,  is it really worth it? As a result the team has conducted extensive interviews with four AI projects to stress-test their method and produce reports on their conduct - specifically asking if their actions are in the public interest and sustainable.

We have been parachuted in to code these interviews and other qualitative data, prior to analysis, provide a fresh set of eyes on their method (coming from our own disciplines) and to lend a helping hand with their english grammar (a task perfect for two over-eager pedants).

The first week has primarily involved getting to grips with the MAXQDA coding software, introductions with colleagues and researchers and catching up on their prior work and methodology. Nonetheless we feel proud to have already helped proofread a paper that will be presented at a conference, make some suggestions about the method and get stuck in sifting through the large amount of data. In general we do not like bullsh*t (whats the scholars network policy on langauge I wonder), meant in the academic sense (Frankfurt, 1986), and Impact AI is in many ways the bullsh*t busters. As a result we are very taken with the project.

Whilst everything has gone swimmingly (aside from having to spend 3 nights in a Generator Inn hostel), there may have been some room for improvement. A tad more preparation, some German phrases, and a bit more focus would not have gone amiss. We have been proud of our confidence and ambition, as well as what I consider a good balance between humility and challenging our colleagues. However, we must ensure to stay driven, and not get too big for our Berlin boots! Ethically we have been very mindful, as the project itself is so reflective. Its nature keeps the ethical spotlight on the projects it audits, its own methodology, and ourselves.

Having moved in to our apartment, immediately adopting the conversational style of a married couple, we are now settled and domestic. This, and Ben’s homemade hot sauce, should fuel us for the coming week. Focusing next week on staying collaborative, intentional and transcendent, we hope to come in to the office raring to go on Monday, so long as the Berlin weekend doesn't take us out first….