Dispelling doubt in AI with quality data 

Jay Limburn at Ataccama explains the vital significance of data quality and data governance for successful implementation of artificial intelligence

 

Businesses that don’t move quickly to adopt AI will struggle to stay competitive in the near future. However, the path to successful AI implementation is littered with failed projects, low internal adoption, and poor-quality data. It’s no wonder that over half of data leaders today report that improving data quality and accuracy is their top priority for 2025. 

 

This finding from Atacccama’s Data Trust survey highlights the ongoing issue organisations face in becoming AI-ready. Data leaders recognise it as an opportunity and a challenge, with the research showing 72% fear not adopting AI will cause their companies to fail. 

 

The high cost of poor data quality

Leading industry analyst firm Gartner has calculated that organisations lose an average of $12.9 million annually due to low data quality, highlighting the need to tackle this institutional challenge urgently. Yet, despite the business case to be made, only 3 per cent of organisations have taken the necessary steps to ensure their data quality is up to standards.

 

The consequences of poor data quality are far-reaching, impacting almost every aspect of a business. Data-driven insights are important for accurate business decision-making, to direct product innovation, and to enhance customer experience. When data is incomplete, outdated, or inaccurate, business users cannot trust it, resulting in inefficient operations, increased risk and missed opportunities.

 

Adding AI and generative AI technologies into the mix adds to this complexity. These are evolving quickly and enhancing functionality with chat, search and early reasoning capabilities. 

 

The intersection of data governance and AI regulation

Successful AI starts with data governance, and companies should pursue a strategy that will enable them to trust their data. The Ataccama Data Trust Report shows that one in five organisations still admit to a lack of data and AI governance frameworks, illustrating that many enterprises still lack the necessary foundations for ethical, valuable AI adoption.

 

The EU AI Act, which will come into force on February 2, 2025, sets out guardrails for AI development using a risk-based approach to cultivate safe, transparent and ethical AI. 

 

Set to prohibit AI systems that are deemed to pose an unacceptable risk on February 2, 2025, the Act requires businesses to ensure that their data is of the highest quality to build and deploy AI systems that are ethical, transparent, and trustworthy. 

 

The regulation mandates that businesses meet certain data standards, where data used for AI must be “fit for purpose” and non-discriminatory, to avoid misuse of AI and ensure the safety of the systems they build. This means businesses need to ensure their datasets are comprehensive, representative, and free from bias. 

 

This presents a significant challenge for businesses, as one-third of them are currently not on track to meet the requirements of the EU AI Act by February 2025. This delay in compliance can have far-reaching consequences including a potential fine of up to €30 million and the resulting reputational damage.

 

Optimism in AI’s potential remains

Organisations that can navigate these external forces are positive about the impact AI will have on their business. Done properly with trusted data, 57% of data leaders believe it will transform customer services - this was most anticipated in enterprises at more advanced stages of AI adoption (71%), indicating customer loyalty plays a key role in sustaining business growth. Other areas in which AI will have material impact include directing product and services innovation, and guiding marketing and sales (39%. It’s clear why exploring AI is a data management priority for 54% of data decision makers. 

 

Data is also taking on an increasingly strategic role in sector-specific use cases and planning for improved outcomes. Insurers, for example, view operational efficiency, increased efficiency and productivity, and increased competitive advantage as the main drivers for AI adoption. Given their need to manage underwriting and claims processes, combat fraud, and ensure accurate reporting, AI will only aid ethical, accurate outcomes if it is using quality data. 

 

Transparency and ethics are also common themes in other industries. Healthcare organisations are particularly likely to face AI-related challenges, to begin with, arising from difficulties in integrating new technologies with legacy systems. Nonetheless, their outlook remains positive - 58% view AI as a pivotal tool for predictive analytics, enhancing data analysis and decision-making processes.

 

Given the sensitive nature of the data collected by healthcare settings, data quality is critical to being able to identify what data assets contain and for which purpose they can be used. Not only is this required for responsible, ethical use, it is also key for healthcare providers to be able to use AI tools to enhance decision-making and reduce human error. 

 

The path to data trust 

Building data trust requires transparency, accountability and communication around how data is sourced and handled, with training provided to employees to ensure adherence to internal policies.

 

The first step in improving data quality is to organise and secure all available assets, ensuring that data is accurately collected, stored and maintained across the organisation and for what purposes it is used.

 

Businesses must also focus on data cleansing to remove inaccurate or duplicate data from their systems, and regularly updating their datasets to reflect accurate information. This will help to understand what data they hold, get visibility into its origins and monitor how it changes over time. 

 

Finally, organisations should invest in AI-powered data quality tools that can automatically identify and correct issues in real-time, reducing the burden on employees and ensuring continuous data quality. 

 

As the digital landscape evolves and regulations like the EU AI Act grow stricter, a solid data strategy is no longer optional—it’s critical. Poor data quality doesn’t just cost money; it amplifies risks, invites compliance penalties, and holds back operational efficiency. With AI and generative AI advancing at breakneck speed, defensive tactics to dodge fines won’t cut it anymore. The businesses winning will be those who embrace data quality and AI as strategic assets, leaving cautious competitors in the dust. 

 

---

 

Jay Limburn is Chief Product Officer at Ataccama

 

Main image courtesy of iStockPhoto.com and sankai