Australian Institute for Machine Learning

Name (Individual/Organisation)

Australian Institute for Machine Learning

Responses

Q1. How could the purpose in the ARC Act be revised to reflect the current and future role of the ARC?

For example, should the ARC Act be amended to specify in legislation:
(a) the scope of research funding supported by the ARC
(b) the balance of Discovery and Linkage research programs
(c) the role of the ARC in actively shaping the research landscape in Australia
(d) any other functions?

If so, what scope, functions and role?

If not, please suggest alternative ways to clarify and define these functions.

There is significant ambiguity about the eligibility of AI in health projects for ARC funding under the ARC medical policy. Under a strict interpretation of the policy (as has been applied at times), these projects are deemed ineligible for ARC funding, whilst also failing to satisfy the eligibility requirements of NHMRC funding, due to the computer science nature of the projects. This leaves a problematic gap in Australia’s funding system that is seeing important areas of research completely ineligible for funding in Australia.
We propose that in these cases rather than a grant simply being deemed ineligible for funding under the ARC, there could be an explicit limitation on how ARC-sourced funds can be expended. For example, a limitation could be applied to deem clinical, animal and human research trials ineligible for ARC funds.

We also note that AI projects have failed to receive support under the centre of excellence scheme since 2014, and in this year’s round of laureate fellowships only 1 out of 40 applications in the engineering, information and computing sciences category was successful, with none in AI - a technology on the List of Critical Technologies in the National Interest. We would suggest the ARC reviews its assessment of computer sciences projects to ensure the metrics of success recognised internationally in these fields are being compared appropriately with other fields. We note that in the critical technologies’ profiles supporting the recent consultation process on the List of Critical Technologies in the National Interest, a measure of research impact was used (for AI and ML) that doesn’t reflect the international rankings recognised broadly by the field.

Q7. What improvements could be made:

(a) to ARC processes to promote excellence, improve agility, and better facilitate globally collaborative research and partnerships while maintaining rigour, excellence and peer review at an international standard?

(b) to the ARC Act to give effect to these process improvements, or do you suggest other means?

Please include examples of success or best practice from other countries or communities if you have direct experience of these.

The ARC Linkage grant application process is too slow for technology-related projects. In practice, the total time taken for the process from conception until the outcome, is approximately 12 months, with three months to put the application together, and between 6-9 months for the outcome to be announced. Technology-related fields require a much shorter timeframe to remain competitive within such a fast-paced sector.

We suggest that a new technology linkage stream be created with a more condensed application and assessment process to a maximum of 3 months in total.

Submission received

14 December 2022

Publishing statement

Yes, I would like my submission to be published and my name and/or the name of the organisation to be published alongside the submission. Your submission will need to meet government accessibility requirements.