War of words on the AI front
Briefing stoush: Ed Husic and Karen Andrews in a dispute over artificial intelligence
As if anyone needed reminding that a federal election looms, a war of words has broken out between the offices of Industry Minister Karen Andrews and shadow human services minister Ed Husic over a briefing on, of all things, artificial intelligence.
Late last year, Mr Husic approached Ms Andrews’ office seeking a briefing on the progress of an AI technology roadmap report being prepared by the CSIRO unit Data61 and the Department of Industry, and to get an understanding of the thinking in the report.
The request was knocked by the Minister’s office – not once but repeatedly – according to Ed Husic and he is not happy about it. These briefings are quite routine and rarely rejected, he says.
While there are no specific rules around such briefings, by convention they are commonplace – although the understanding is that they are done in the background, quietly and without any resulting overtly politicisation.
Even in the hyper-partisan times we live in, governments see merit in ensuring both the government and opposition benches the opportunity to understand the detail of evolving policy – particularly where there is complexity.
The government ear-marked $29.9 million toward improving Australia’s capability in artificial intelligence and machine learning. While the lion’s share of this funding was to have been distributed through AI and machine learning projects with the Cooperative Research Centre program, about $3 million was directed to Data61 to produce an AI technology roadmap and an AI Ethics Framework.
Data61 says it is still working on a first draft with the Industry department of both the AI technology roadmap and the Ethics Framework. The work is incomplete, and matters of timing (about when the reports would be completed and when they might be publicly released) are for the minister’s office.
And this is where crankiness has set in. Mr Husic says it was not necessary that final versions of the reports be finished. He says he was simply seeking [through the minister’s office] a generic briefing from Data61 on the current thinking on AI, and that nothing from the briefing would be made public or disclosed in any way.
“I have had briefings [via] a number of ministers, and they’ve been great in making those meetings happen,” Mr Husic told InnovationAus.com.
“This is the first time in a long time that I can remember where I have been refused a pretty straight-forward briefing.”
Having travelled to Japan over the summer break for briefings from Japanese government departments on AI technology and ethical frameworks for AI systems, Mr Husic says it’s galling that he can’t get a briefing from the Australian government.
As the chair of the G20 summit to be held in June, Japan has identified artificial intelligence as a key agenda item for discussion – particularly on international rule-setting on ethics issues.
“And so I have had really very detailed briefings from different government departments [in Japan] on what the Japanese are doing. They’re quite open with me about their planning and I met with a number of Japanese companies [working in AI] as well,” Mr Husic said.
“Yet when I have tried to meet with our own government about where we are on the AI, I haven’t been knocked back just once, I have been knocked back repeatedly.”
For its part, Karen Andrews’ office simply says that in the absence of the completed reports that were funded in the 2018-19 budget
“Mr Husic was informed at the time of his request on 6 December that the Government document was not yet finalised, and when more information was available his office would be contacted,” a spokeswoman for Minister Andrews told InnovationAus.com.
“The document is currently under development by Data61 and the Department of Industry, Innovation and Science,” the spokeswoman said.
Which brings us to the timing of the release of these AI documents. It is understood the Artificial Intelligence Technology Roadmap report should be completed and released to the public within four to six weeks.
This would mean that we might see it before the election, but also that we might not. Perhaps it becomes a pre-election announceable.
The production of an AI Ethical Framework will likely take longer. It is understood that a paper will be published, to be followed by a consultation period to give the public a chance to understand what is being proposed.