On March 14, 2019, the Defense Innovation Board held a listening session on Artificial Intelligence Principles at Carnegie Mellon. The DIB is an independent committee made up mainly of tech execs and academics that advises the DOD. I wasn't planning on giving a comment at the session, but I got so angry listening to the pre-written publicity statements read aloud by defense contracting tech companies, that I couldn't stay seated.
Here's my very last-minute statement:
Hello, my name is Maggie Oates. I'm a PhD student here in Societal Computing and I work in Cylab, which is the cybersecurity lab here.
Let me first say that I disagree with the very enterprise and existence of the Defense Innovation Board, as it seems like a tool to lend credibility to the project of advancing military efficiency and further escalating the baseline of 'defense.' Second, I disagree with CMU's continued involvement with the military and am hard-pressed to think of an ethical and responsible use of AI at all [in defense].
That said, I would like to focus on something else today and that is what I view as an externality of the project of AI in the DOD. And that is the growth of civil surveillance, both domestically and abroad. The development of machine learning algorithms relies on massive amounts of data, of course. And while methods are being developed to reduce the amount of data required, or to reduce the amount of labeled data required, these methods often correlate with having the downside of being hard to explain and more difficult to verify, making them an unlikely use in the DOD's context. Beyond that, the project of labeling data often rests on exploitative labor practices. So I stand here to assert that any responsible principles must address the effects that DOD AI will have on surveillance, not only from the state, but also from the tech companies that will be the first line in building that AI. This topic is absolutely not out of scope. Thank you.
You can view the video and comments on the DIB site.
