An ANU study has found that just 17 per cent of people think AI will create more jobs than they eliminate.
It’s not that there are lots of Luddites “it is the speed that feels different, as well as the cumulative complexity,” says Nicholas Biddle (ANU Centre for Social Research and Methods).
“Australians still think that AI will have a positive effect but there is far less support for AI having a positive effect compared to solar or wind energy, or vaccines,” Professor Biddle writes in a new report on an ANU poll, taken in April.
The ambivalence is important – while community understanding of how AI works may not matter much, people “do need to have sufficient information to be able to balance the trade-off between costs and benefits,: he says.
The same applies to other techs where people hold strong opinions, nuclear tech, nanotechnology, biotechnology, and different forms of green energy.
But AI raises more red flags than green. Only 20 per cent of the poll was “more excited than concerned” at what it could do.
MPs are reading the national mood – with parliamentary inquiries on AI announced in the Commonwealth, NSW and South Australia parliaments.
Education Minister Jason Clare has asked the House of Reps Standing Committee on Employment, Education and Training to inquire into “issues and opportunities presented by generative AI” from childcare to HE.
In contrast, the NSW Legislative Council has commissioned an inquiry with a focus on employment and community protection. “Failure to plan and to regulate is simply not an option,” committee chair Jeremy Buckingham says.
And in South Australia, a House of Assembly committee will focus on what’s in AI for SA. There is a free kick for university merger advocates in the terms of reference which include, “development of a strong AI research and development sector, the attraction of international AI investment, and the training and retention of AI talent.”
All three committees are just beginning, but submissions should not be long in arriving – what with terms of reference being dead easy for GPT-4 to address.