Australian Digital Alliance

Our Chair @derek295w is closing out the Forum now. #ADAForum

.@sebchan: The @ACMI ustilises machine learning to enable 'beyond human scale' access. #ADAForum

.@sebchan: Culture reduce to data and data is now raw material for pattern recognition. #ADAForum

Miller: @spawning_ created the 'Do Not Train' tools suite to help creators mark their work using machine-readable code. #ADAForum

Cullen Miller: Rights holders didn't have a sufficient way to signal and register their preferences for how their data could be used in training data. This is complex. Plus what use is a register if it is not enforceable or respected by developers? #ADAForum @spawning_

Lanchester: Innovation is not the sacred ground of AI developers. Our creative industries are innovative. Innovation is shared across many industries. Licensing has facilitated innovation for decades. #ADAForum

Lanchester: 'Exempt me from the law' is a strategy of the powerful. It is a claim to the labour of others for free. #ADAForum

Lanchester: Australian authors are now well positioned to push for licences, especially against the developers of foundation models. They face resources imbalance and an information asymmetry. #ADAForum @asauthors

Lanchester: We need to set twin goals: one to support ethical AI development so we can realise its promise. We also need to support Australia cultural and creative life. #ADAForum

Olivia Lanchester: What we see at the @asauthors is wrtiers struggling to make a living at the end of two decades of decreasing incomes. #ADAForum

Olivia Lanchester: Authors are excited by AI as a new tool. They are also concerned about AI being a vehicle through which their paid opportunities may be deminished or supplanted. #ADAForum

.@spamthemusical: AI is a combined intelligence between man and machine. #ADAForum

.@spamthemusical: Now when I work with AI I work with training data that is so big that I don't even know what it consists of. It is like an anonymous choir and I am the conductor. I need to connect myself to this training data to create some kind of harmony out of it. #ADAForum

We end the Forum with a panel looking at challenges and opportunities for AI with @spamthemusical, Olivia Lanchester, Cullen Miller and @sebchan. #ADAForum

Interesting- for collective licensing to be even considered be a possibility the amount and depth of data disclosure and sharing required to even attempt to link outputs with original works and creators and owners (if that were possible) to enable distribution to creators and owners would be wide ranging. That would mean a level of transparency and disclosure by AI entities which isn’t possible - really - I might have a better chance as a citizen of India being selected (not elected) as the President of the United States of America than this level of disclosure ever becoming a reality.

Collective licensing for AI training is a fantasy. The reality will be an AI tax used to fund worthy causes. That might be a good idea, but it won't be like ASCAP

.@robynVDyk: When the Memorial published a publicaiton a few years ago the publisher treated three words in a row as a quote and each quote needed a clearance for use. No publisher is taking risks on quotation but they create significant labour to clear. #ADAForum

.@robynVDyk: We have been using s 200AB at the War Memorial for 15 years and we have not had one issue, not one notice-and-takedown. There is a catastrophising that just doesn eventuate. #ADAForum

Driscoll: We find that third-parties are reluctant to use orphan works. Organisations can rely on s 200AB but end users such as documentary filmmakers can't and we are passing the risk to them when we provide content. #ADAForum