And as the uptake grows, it’s also having a direct impact on how companies engage with the Research and Development Tax Incentive (RDTI). We’re taking a look at this in more detail, and remember at any point if it all gets too confusing – don’t worry! We have an expert here at Standard Ledger who loves tackling complex tax claims, so book a time to chat with John Nixon to get up to speed.

How AI aligns with the RDTI

AI related projects usually align well with the key eligibility criteria of the RDTI, which focuses on systematic experimentation and the generation of new knowledge:

• Advancement in technology: Companies that are pushing the boundaries of what’s possible with AI may be creating new algorithms, models, or applications that require experimentation to develo
• Technical uncertainty: Developing custom AI solutions sometimes involve unpredictable outcomes and problem-solving that are not always achievable through existing publicly available knowledge, or tools
• Systematic experimentation: Training, testing, and refining AI models requires iterative processes, that can qualify as experimental activities.

Industries leveraging AI for R&D

There’s a few sectors in Australia that are actively and effectively incorporating AI into their R&D strategies. A few examples of which industries, and what they are doing with AI are:

• Manufacturing: Predictive maintenance, quality control, and robotics
• Healthcare: Diagnostic tools, patient monitoring, and personalised treatment plans
• Agriculture: Machine learning for crop prediction, yield optimisation, and pest control
• Financial services: Fraud detection, risk assessment, and automated compliance systems

Challenges and increased scrutiny

Despite its undeniable potential, AI-related RDTI claims are attracting greater scrutiny from the Australian Taxation Office (ATO) and the Department of Industry, Science and Resources (DISR). While it’s still relatively ‘new’ for some industries, a few reasons why it might be put under the microscope are:

• Blurred lines: It can be difficult to separate ‘routine software development’ from genuine R&D activities in AI-based projects
• Use of existing platforms: Many AI projects rely purely on adopting or adapting existing third-party tools and this may not meet the threshold for generating new knowledge
• Poor documentation: Companies sometimes fail to record detailed technical documentation that evidences how their AI efforts meet the definitions of R&D

Best practices for claiming AI R&D

To maximise the likelihood of a successful RDTI claim, it’s really important to focus on:

• Keeping comprehensive technical records of hypotheses, experiments, failures, and iterations from the start
• Demonstrating technical uncertainty, and why the outcomes could not be known in advance
• Clearly distinguishing core R&D from any supporting or commercial work that you do
• Seeking advice from R&D tax specialists (like us!) to validate eligibility before lodging a claim, as this will make sure you are going to provide the right information

What’s Next

AI is there to take advantage of – so don’t miss out just because it’s new and different. Remember, spellcheck was new once too! AI is playing an increasingly vital role in Australia’s innovation landscape, as it offers new opportunities for businesses like yours to benefit from the R&D Tax Incentive. But even though developments using the technology will often meet the RDTI eligibility criteria, the road of navigating the complexities of the RDTI needs careful planning, strong documentation, and a clear focus on experimentation/new knowledge creation. As the regulatory environment evolves to keep up, we all need to make sure we stay informed and proactive so that AI projects not only drive value, but also qualify for the government support they deserve.

As always – get in touch to discuss any stage of your RDTI projects, or check out our resources page for a ton of info on startup to scaleup financial matters. We’d love to help, and we know what you’re going through (we’ve been there too!).