There isn’t a person reading this who hasn’t experienced the fallout of poor data practices and governance. Recently I spent many frustrating and fruitless phone calls trying to unpick a series of fiddly, yet entirely preventable customer service issues with a well-known retailer that stemmed from the simple duplication of my customer record. I had until then been a loyal customer, perhaps a little too loyal as I was a member of their exclusive VIP programme (reserved for only the very biggest spenders…). But having experienced first-hand their lack of care for my data, not to mention their fairly hopeless customer service, I vowed never to step through their doors again; and made sure I told anyone around me who would listen about their shoddy data practices.

Organisations have been saying that data is their greatest concern since the dawn of the digital age. It was way back in 2006 that British mathematician Clive Humby referred to data as ‘the new oil’; so why are many organisations still grappling with the basics? And what does this mean for organisations keen to leverage the latest AI tools? Will these fast-moving developments exacerbate existing data concerns for businesses, or worse, create new ones?

In our June ‘Insights with Impact’ poll, we set out to understand what members’ top data concerns are in this digital-first / AI world. Our poll was conducted in the context of ongoing media debate about data ownership and privacy, not to mention high-profile cyber-security threats including the UK armed forces data hack in May. We posed the statement:

My greatest concern about my data is:

• Quality
• Accessibility
• Bias and distortion
• Security, privacy and resilience

The top response was data ‘quality’, which encompasses data accuracy, consistency, validity, completeness, uniqueness, and timeliness. We saw from my experience above the impact that poor data quality can have on customer trust and a brand’s reputation, as one small data quality mistake can have widespread and unforeseen repercussions. Arguably data quality is more important than ever in the age of AI. If the quality of data used to train an AI model is incorrect, incomplete, inconsistent, or biased the AI program will generate biased, incorrect, incomplete, or inconsistent results. So for good-quality AI outcomes, you need good-quality data. Conversely, high-quality data will generate more accurate predictions and more personalised recommendations. We spoke with Nic Granger, Director of Corporate, North Sea Transition Authority who selected ‘data quality’ in the poll.

“Quality data is crucial to every organisation. Better data leads to better decisions. Making decisions using poor quality data can risk making the wrong decision, so as technologies such as AI become more mainstream we all need to take a step back and think about the quality of the data that fuels them.”

Unsurprisingly, ‘security, privacy and resilience’ was the next highest response in our poll. Every day seems to herald a new concern in the media about AI privacy, whether it’s the comments from Elon Musk regarding Apple’s new AI tools being a security risk, to concerns about Meta training its AI model on European Facebook users’ data. We spoke with a respondent from a global business services provider who selected this option and he told us:

“It’s important not to rush in. Security and privacy issues around AI must be paramount if you’re going to win the trust of stakeholders and, above all, protect your data. However much you want to advance your AI initiatives, it would be unwise to entrust your data to an LLM provider; that’s your IP and it’s too precious. A risk-aware approach is to build a secure environment where you can try out AI applications, and carry out robust testing so you can show your stakeholders - your users, your customers, and your partners – that you’re responsibly deploying this technology.”

‘Bias and distortion’ was a surprisingly low data priority in our poll, when considering the rise of AI and concerns about the potential dangers of AI fabricated images or statements. With more than 50 countries preparing for elections this year, one might have expected the results to reflect the very real concerns many have about how AI is being used in electoral campaigning. Data bias is also becoming a greater concern for many organisations and there have been numerous high-profile examples of AI tools amplifying bias in gender and race. Data bias stems from both discriminatory data and algorithms that are baked into AI models, which makes it very difficult for organisations to eliminate.

It is clear from these results that although many of the fundamental concerns around data remain the same as maybe a decade ago, the rise of AI provides a powerful incentive to get to grips with these challenges now. If you are interested in finding out more about how to leverage AI, why not join us at Unlocking ROI from Technologies Driving Transformation and Innovation on Wednesday 17 July? Also, look out for next month’s Insights with Impact where we will be exploring the top use cases for AI according to our latest poll.