Michael Veale

In relation to specific PhD projects, I don't think the question is wheter FB defines direction or not, but whether it institutionally constrains it. Contractual terms between UCL and Facebook on freedom to publish and IP ownership should be published.https://twitter.com/riedelcastro/status/1364526667427958785 …

Industry funding or co-funding is common (through eg CASE), but it's not common for companies to run a structured PhD programme. This isn't about compromised individual researchers; it's about institutional reliance of universities on private companies.https://twitter.com/emilianoucl/status/1364566724000022529 …

Some questions: 1) Is UCL publishing the total value of direct and in-kind (e.g. compute) resource it is being provided by Facebook? 2) Does UCL disclose how many of its academics are either funded by FB or on part-time FTE contracts jointly with FB?

3) Are students in practice expected to sign NDAs to undertake their PhD? 4) Has UCL negotiated any actual contractual terms with FB that protect its students' indepencence and freedom to publish? What are they?

5) Are papers written by the student subject internal FB review which also considers business interests (e.g. the recent Google @timnitGebru situation) — even de facto by virtue of being co-authored with a FB supervisor?(students can hardly refuse to add a supervisor's name)

It's unclear why UCL has to do this. I can see the pressure if it was a field with little PhD funding, but public funding for AI PhD students at UK universities is not difficult to secure (following consequences of report by @DameWendyDBE). Supervision capacity is the limiter.

The sum for this programme does not seem very big, either. 1m over 4 years. A cheap price for a company that routinely operates outside the law to legitimise itself by aligning even more closely and explicitly (there's even a press release!) with a top university.