4 in 10 executives don't trust their companies' data to generate accurate AI outputs
A new survey of executives and AI leaders finds that while business decision makers believe in the potential of AI, many lack confidence in their companies' strategy execution and data preparation to ensure trust in AI outputs. Additionally, 7 in 10 executives say their companies' AI strategies are not fully aligned with their current business strategies.
The study was conducted for Teradata by NewtonX, a leading global B2B market research firm, and included expert interviews and quantitative surveys of executives and decision makers with insider knowledge of their companies' AI strategies and execution. All surveyed are responsible for or use AI in their roles. 61 percent said they have complete confidence in the reliability and validity of AI output, while 40 percent believe their company's data is not yet ready to deliver accurate results.
“The foundation of AI is clean, reliable, and trustworthy data, because it's the backbone of AI output,” said Jacqueline Woods, Teradata's chief marketing officer. “Achieving complete trust remains a challenge for many executives, but our research shows a growing understanding of how to arrive at trustworthy AI at enterprise scale, confirming that Teradata is well-positioned to help customers with these business goals.”
AI is essential, but a clear, aligned strategy is lacking
While 89% of enterprise executives believe AI is necessary to stay competitive, only 56% say their organization has a clear AI strategy, and only 28% believe their organization's AI strategy is closely aligned with and supports broader business goals. Most successful AI implementations are happening at the departmental level, with only 12% deploying AI solutions across the enterprise and 39% implementing AI within a specific department.
Executives cite significant productivity gains (51%) and improved customer experience (50%) as the most important benefits of AI. However, despite the potential for customer-facing applications, most executives prefer to undertake AI projects that enhance internal processes, as these projects tend to minimize AI risk and are perceived as more likely to improve cost control than drive growth.
- Nearly half of the executives surveyed are successfully using AI to improve employee productivity and collaboration (54%) and support decision-making (50%), but only one-third are using AI for product development (30%) and sales and revenue forecasting (30%).
- More than half (57%) of executives surveyed said they are concerned about how AI failures could impact customer satisfaction, company reputation, or both, noting that to make AI successful, there needs to be better alignment between AI and business planning.
- Even for internal projects, 63% of executives surveyed reported using a combination of closed and public data sets, while only 29% rely exclusively on closed data sets.
- Barriers to effectively scaling AI projects include:
- Shortage of AI technical talent (39 percent)
- Lack of budget needed to scale AI projects (34 percent)
- Difficulty in measuring business impact (32%)
- and lack of technology infrastructure (32 percent).
While 73 percent of those surveyed consider their company to be an early adopter of many technologies, 60 percent said their company's level of AI adoption is “on par” with competitors, and only 27 percent consider their company to be leading their industry in AI adoption.
Increasing trust is an obligation
For executives, trust in AI projects and their results is crucial. One participant stated, “… we want to be clear to our customers about what data was used to train the model,” noting that choosing the wrong training set can easily introduce bias into the model. Another participant stated, “… master data management is not sexy, but… if you base everything on data and that data is flawed, you run into problems.”
Beyond unbiased data, survey participants said operational efficiencies (74%), demonstrating successful use cases (74%) and improving decision-making processes (57%) are the top drivers of trust within their organizations for new AI implementations. Additionally, critical to trust in AI is prioritizing vendors and partners that facilitate seamless integration with best-in-class AI solutions (67%).
Among other findings, survey participants noted:
- Reliable and validated results (52%), consistency/reproducibility of results (45%), and the brand of the company building the AI (35%) are the three most important factors in building trust in AI.
- Important aspects of trustworthy AI were cited as security (61%), transparency (55%), governance (45%) and improved business performance (40%).
Contributing to AI success
Respondents cited a clear strategic vision and leadership support (46%), effectively communicating the benefits of AI to stakeholders (46%), and sufficient investment in AI technology and infrastructure (41%) as key drivers of their AI success to date.
The majority of respondents (84%) expect to see results from their AI projects within a year of implementation, and more than half (58%) say they can quantify results within six months. Additionally, 60% say they are already seeing “demonstrable ROI” with their existing AI solutions.
“There's a huge opportunity to increase trust in AI through better alignment of business and AI planning, but planning alone can only go so far,” says Woods. “Working with the right partners and solutions can help you quickly show accurate results and ROI from your AI projects, accelerating trust. But remember, every successful AI project starts with a foundation of clean, trusted data (what I call 'golden data') based on solid data sets and providing full transparency, and this is where Teradata helps.”
About the survey/method
The survey was conducted in the U.S., Europe, the U.K. and Asia among C-suite executives and AI decision makers from companies with more than 1,000 employees and annual revenues exceeding $750 million. The survey reached nearly 300 AI executives from companies such as Nike, P&G, Hermes Paris, Allianz Partners, Prudential Financial, Honeywell and Novartis, with roughly half of the respondents based in the U.S.
Sign up for the free insideBIGDATA newsletter.
Join us on Twitter: https://twitter.com/InsideBigData1
Join us on LinkedIn: https://www.linkedin.com/company/insidebigdata/
Join us on Facebook: https://www.facebook.com/insideBIGDATANOW
