Balancing Innovation and Incentives

One of the more notable revelations from the Australian Government’s 2023-24 Mid-Year Economic and Fiscal Outlook (MYEFO) brings light to a $2.1 billion expected increase in R&D Tax Incentive (RDTI) costs over four years.

A significant part of this surge can be attributed to rapid advancements in Artificial Intelligence (AI) and machine learning (ML) over the last few years. This trend prompts a crucial question: as AI and ML becomes more widespread, could their growing ubiquity potentially impact or reduce eligibility for R&D incentives?

It’s a complex balancing act, promoting cutting-edge research while ensuring the RDTI remains targeted at truly innovative work. The rise in software claims, from 30% of RDTI applications in FY19-20 to over 40% in FY21-22, further illustrates this rapid technological evolution. This comes at a time when Australia’s R&D spending as a percentage of GDP, according to data from FY21-22, is at a low of 1.68%, contrasting against the OECD average of 2.5% in 2019. This is despite the government previously committing to elevate R&D spend to 3% of GDP.

Australian treasurer, Jim Chalmers, in discussions with InnovationAus, hinted at a broader review of tech-related arrangements to maximise the benefits of technological adaptation and adoption. But, as these changes come into effect, could we see RDTI eligibility reform happening soon for AI and ML?

Through the lens of RDTI experts here at Sprout, this reform could involve a more nuanced approach to defining innovation. This may include setting stricter guidelines to differentiate between routine software development and genuinely ground-breaking AI research. These changes would aim to ensure that the RDTI incentivises truly innovative and transformative projects in these rapidly evolving sectors.

If you are considering making an application for the R&D Tax Incentive for a project that involves AI and/or ML specifically, some of the eligibility factors you should consider are:

1. Is there at least one member of the R&D team who has a background or expertise in AI technology and machine learning techniques? Or, have you engaged with professionals in this field? Eligible applicants are expected to have engaged and utilised expert, professional knowledge to help determine that the project objectives are indeed innovative and contain technical uncertainties.

2. Are you building an API or an app that communicates with a Large Language Model (LLM) without building an LLM (or similar technology) yourself? These types of APIs are not necessarily technically innovative, and leveraging existing AI technologies made by other companies may trigger the “Core Technology” clause or be looked upon unfavourably in an audit/review.

3. Can you demonstrate and explain why your project requires AI, or more specifically ML, to achieve the project objectives? AI may seem like a guaranteed method of improving your technology, but in many cases there are existing techniques or methodologies (e.g. statistical) that can achieve the desired result without the use of AI.

For more guidance and assistance, reach out to Sprout Scientific for a complimentary consultation.