When I arrived at ACMI three years ago, I had a relatively naïve understanding of how much museums invest in the quality and robustness of their digital technology. Obviously, no-one wants their technology to fail, but museums — especially technology-rich ones like ACMI — face challenges of robustness, demand and scale that are unlike any other industry sector: How can we plan for no accessible cabling but easy maintainability? An artist has made a work with technology we’ve never encountered before.

Artist’s impression of technology doing its thing in our new exhibition As Lead Engineer and project CTO, I have headed up the multi-year technology design and implementation for ACMI’s $40m Re/newal. One of the guiding principles of the renewed ACMI is “curation by humans, enabled by technology” (sigh, I guess my top secret curate-o-bot AI can wait for another day). We wanted to build a system that curators can use to choose and rapidly deploy playlists of content around ACMI’s galleries, whether the media comes from ACMI’s Collection, edit suite, special exhibition videos or elsewhere.

This is just a quick post to say hello, we made an exhibition-scale internet-of-things media player, and we’ve open-sourced it so you can use it too. XOS media players playing through various kinds of demo content on various screens as part of our long-term testing — thanks to sighmon for the photo Why did we roll our own media player? As the world’s most-visited museum of the moving image, ACMI has a lot of, well, moving image to show.

Last week the British Museum wrote about how they were making sense of their TripAdvisor reviews, using natural language processing techniques and data visualisation — work they presented at MCNx. It was good-looking (if dense) stuff, which surfaced new insights into what visitors found important about their experience, and pointed towards some root causes. But I couldn’t help thinking that the approach suffered from throwing away the ratings data, and was very resource-intensive— the article concludes with plans for two PhD data science interns and a partnership with the Alan Turing Institute to continue the work.

My first few weeks at ACMI have been a whirlwind of discovery, exploration, planning and excitement around its forthcoming renewal. One of my projects is to continue to flesh out explorations of machine learning and related AI techniques to ACMI’s collection of Australian moving image works. This is not that project. Or if it is, then it’s more of a fun steam-letting-off part of that project. Drawing from Dan Hon’s experiment in generating British place names, I trained a Recurrent Neural Network (RNN) on titles from ACMI’s collection, and asked it to generate some more.