Deborah Henderson | AI in 2024: Managing risks and charting the future

As a QUT eMBA alumnus, Deborah Henderson brings her wealth of knowledge and ethical acumen as guest editor of the February 2024 edition of the QUT Alumni newsletter. Deborah delves into the profound impact of AI on businesses, emphasisng the opportunities, risks, and pivotal role of ethical considerations. A Data and AI Partner at KPMG Australia, she advocates for a careful and risk-aware approach to AI development, offering her views on five things organisations need to consider as they navigate their AI journey. 

AI has been around for a long time. I’m probably going to ruin the magic a little by saying that in its fundamental form, it is nothing more than mathematics, software development and data. So why the hype now?

With computing power increasing, data storage costs reducing and the way we interact with this technology becoming more natural it is no longer limited to those with specialised skills.

A KPMG survey on generative AI found that more than three quarters (77%) of executives see this software as having the largest impact on their business out of all emerging technologies. Another 73% of respondents believe it will increase workforce productivity and 71% plan to implement their first generative AI solution within the next two years.

But it would be remiss to talk about the opportunities of AI without talking about the threats and risks. In the rapidly evolving landscape of AI, legal and ethical considerations are gaining prominence. The rise in legal cases surrounding data used to train generative AI models underscores the need for robust regulatory frameworks. I believe the solution is in developing Large Language Models (LLMs) based on data within our control.  At KPMG Australia we’ve been very careful in taking an ethical and risk-aware approach to developing our AI capabilities. For example, KPMG’s private AI Assistant called KymChat offers a powerful internal policy navigator that ingested KPMG’s 900+ page quality risk manual to answer questions from staff. The capabilities of tools like KymChat are continuously evolving as the technology improves and risks are better understood and mitigated.

I’m also of the opinion that there are a series of guidelines organisations will need to consider as they navigate their AI journey. These include being able to:

1. Strive for a balance between innovation and risk by identifying opportunities where AI can solve problems and provide benefits that clearly outweigh the risks (e.g. teacher workforce shortage) ensuring it is safe to use
2. Co-design solutions with stakeholders to make sure it meets their needs and mitigates risks like bias
3. Prioritise data governance initiatives and get your data ready to avoid errors being amplified when fed into the AI model (this includes considerations  around data architecture, data platform, data modelling, data security and data quality)
4. Validate the accuracy of GenAI output (it is entirely possible for GenAI models to produce factually incorrect information) and recognise that domain expertise and critical thinking is required to verify that the response from the GenAI model is factual
5. Increase education, awareness, and training around AI (e.g. plan for talent and workforce upskilling, build understanding of potential use cases and knowledge in common AI platforms)

KPMG’s recently published Global Trust in AI report reveals that the current safeguards for governing AI are struggling to keep pace with expectations and technological advancements. While 70% of people expect AI to be regulated, only 25% believe existing regulations are sufficient. Moreover, half of the population admits to not understanding how AI is used. The path forward involves substantial upskilling and training efforts to enable workforces to use AI safely.

Overall, it's clear there is a preference for AI to function as a support tool, with humans retaining equal or greater control. This aligns with the broader sentiment that AI should complement human capabilities rather than replace them.

In the February 2024 edition of the QUT Alumni newsletter (of which I am guest editor), you will find Q&A sessions and articles from fellow alumni offering their own perspectives on the challenges and opportunities presented by AI in various industries. I encourage you to engage with these insights and contribute to the ongoing conversation.

Thank you for being an integral part of our alumni community, and I look forward to witnessing the impact of our collective intelligence on the future of AI.

Deborah Henderson

QUT degree -  Executive Master of Business Administration (2020)

Do you have a question for Deborah? You can connect with her on LinkedIn.