Executive enthusiasm for artificial intelligence is palpable. Our research of 550 companies shows that 97% are dedicating an average of 13% of annual revenue to AI and machine learning initiatives. Unfortunately, that money is buying more than just the tech for AI — it may be funding overconfidence in the C-suite. 

Though 24% of companies report being in advanced stages of AI development, that number changes based on who’s responding. Thirty percent of non-technical executives see their companies as advanced, but only 22% of technical executives feel the same. Our study included 369 technical executives who oversee IT or Data Science/Analytics departments. The 181 remaining non-technical executives oversee business, design, logistics, fraud and other areas.

The confidence gap between the two sets of executives has significant implications for a company’s AI journey and outcomes. Higher confidence numbers come from more senior executives, who are often distanced from technical implementation details. They are more likely to receive favorable updates that gloss over the complexities and challenges of implementing AI effectively.

While having diverse opinions on AI trustworthiness can be beneficial from a strategic standpoint, overconfidence could lead to a simplistic approach that overlooks the importance of high-quality data, increasing the risk of failure and costly mistakes amounting to hundreds of millions of dollars.

[CTA_MODULE]

The crucial role of a strong data foundation

The misconception that AI success is a simple matter of implementation can lead to significant challenges:

• Misallocated resources: With attention excessively focused on advanced AI technologies, technical teams may neglect addressing foundational data quality and integration issues.
• Unrealistic timelines: Organizations are more likely to spend time preparing data than actually building and using models.
• Underwhelming outcomes: Even if an AI model is successfully deployed, suboptimal data can limit its effectiveness and ROI, failing to meet expectations.

AI and ML initiatives need robust data pipelines that automatically move and centralize governed data into analytics-ready formats. Neglecting to build this groundwork, while funneling resources into AI programs starved of trustworthy data inputs, will inevitably lead to negative outcomes.

“An enterprise-wide AI rotation requires a strong data foundation. You can’t just jump to the great data foundation. You need to be in the cloud. You’ve got to have modern platforms.”
- Accenture Chair and CEO Julie Sweet (Source)

Of the 550 companies we surveyed, 96% say they’ve encountered obstacles in AI implementation, ranging from lack of leadership to outdated IT infrastructure. The few companies who’ve navigated around these issues credit their strategic investments in data management, such as governance, security and data movement initiatives. Meanwhile, less advanced companies are spending an average of 67% of their time working with and preparing data.

Companies leading in AI are focused on model development and machine-led decision-making, leading to financial returns on AI investments. For the 40% of surveyed companies that are able to measure their ROI, they saw an average of 52% over the past 12 months.

Poor data quality leads to $406 million in losses

As our research indicates, data quality and governance challenges grow more prominent as organizations progress in their AI journeys. Compromised AI performance resulting from suboptimal data generates losses amounting to 6% of an organization's annual global revenue, or $406 million on average based on respondents from organizations with an average global annual revenue of $5.6 billion. Robust data governance practices are essential insurance policies, fostering responsible and effective AI implementation.

While the drive to be an AI leader is understandable, the payoff will be significantly greater for executive teams that make data readiness the first priority. Without this commitment, AI ambitions will inevitably be derailed by poor data quality, missed deadlines and underwhelming results that fail to justify the investment.

[CTA_MODULE]