Top

Business of Data Meets PwC Chief Data Scientist Prof Matt Kuperholz

To celebrate being named one of the world’s 100 most influential data leaders, PwC Chief Data Scientist Matt Kuperholz talks about what’s next for AI, developing the Responsible AI toolkit and the changing role of the CDAO

What were your greatest professional achievements in 2019?

I have spent more than 20 years applying AI based solutions to solve a wide range of my clients’ challenges.

During this time, AI-powered solutions have continually become more powerful and delivered more value. They have also been progressively enabled by other complimentary exponential technologies, such as increases in data, computing power and connectivity.

But in tandem, new types of associated risk have become more relevant to address. These risks are not limited to the performance of AI. They also relate to newer areas of risk, such as ethical, societal, AI security and control risks.

In 2019, together with a few other analytics partners globally, I lead the research, development and deployment of a collection of tools and services to support our clients in managing these new emerging risks. We have called our solution ‘Responsible AI’.

We developed tools which support the appropriate governance of AI, the development and deployment of ethical frameworks relating to AI usage and the testing of algorithms to ensure fairness, understanding of bias, transparency and explainability are accounted for, while ensuring the deployment of AI is robust and secure.

I consider this some of the most important and societally relevant work I have ever done.

How will you build on those achievements over the next 12 months?

The trends that drove the development of the Responsible AI toolkit are only going to continue and increase in impact.

We are witnessing the effects of AI-powered technical solutions that are not fair, cannot be explained and are not sufficiently robust and secure at an accelerating rate. We are also seeing renewed focus on the requirement for companies to appropriately govern and apply robust, societally acceptable ethical frameworks to their use of AI.

In the near future, it is likely that we will see AI-powered solutions becoming more democratized and commoditized, and so more readily available to a wider group of entities who want to use or adopt them.

Regulatory, customer and citizen pressure to use these technologies responsibly will only increase over the next 12 months.

What are the key challenges organisations will face this year when applying data, analytics or AI tools?

In Australia, there is increasing regulatory focus on the operations of various industries. This focus has increased awareness of the fact that sometimes governance procedures and frameworks have not developed in alignment with the use of data, analytics and AI within the business.

I think the key challenges will shift in focus from technological to ethical and societal and require often significant improvements in governance.

Developing and deploying appropriate governance policies and procedures guided by actionable ethical frameworks will be the key to overcoming these challenges.

How do you think the role of the CDAO is changing? And what’s driving these changes?

As companies have become able to use technology to be more focused on their customers’ needs, the role of the Chief Marketing Officer has elevated in relevance within many organisations.

I think the CDAO is on a similar trajectory in terms of increasing relevance. Successful analytically enabled companies will outperform their competitors, based on their increased adoption and application of data and analytics into all facets of business.

In tandem, the levels of accountability for the CDAO to manage risks and ensure the responsible application of these technologies will increase.

How do you think the way organisations use data and analytics will evolve over the course of 2020?

Even in the face of exponential improvements in data, analytics and associated enabling technologies, I am often seeing Hofstadter’s theorem in action. Everything takes longer than you plan for, even when you factor in Hofstadter theorem!

This feels like the ultimate paradox of data and analytics technology. Things are moving so quickly with underlying technologies, but their adoption into everyday business taking longer than I expected.

In 2020, as with every year this century, I expect to see greater integration of data and analytics into all facets of many large organisations’ operations. However, I expect to continue feeling there will always be additional value to be realised by increasing the rate of adoption.