Poor governance and data management practices, not technological limitations, are the primary reasons artificial intelligence initiatives struggle in organisations, according to new research that challenges recent claims about widespread AI project failures.

A survey of more than 200 business and IT leaders by The Data Warehousing Institute (TDWI) found governance issues were cited by 49% of respondents as their greatest AI frustration. This was followed closely by AI hallucinations at 46% and lack of AI literacy at 48%.

The research directly counters headlines from a recent MIT study claiming 95% of AI projects fail. TDWI researchers argue their findings complete the narrative by highlighting organisational rather than purely technical barriers.

"The roadblocks to AI success aren't just technical; they're organisational," said Meighan Berberich, TDWI president. "Companies that put governance, literacy, and a builder mindset at the centre of their strategy won't just survive the AI transition, they'll thrive."

The study distinguished between "consumers" who use packaged AI tools and "builders" who embed AI into workflows with their own data. Builders reported faster decision-making (64%) and increased innovation (46%), while consumers mainly cited time savings.

Despite implementation challenges, 90% of survey respondents already use general-purpose generative AI assistants such as ChatGPT and Perplexity. Nearly half of builder organisations are experimenting with workflow automation including onboarding systems and incident reporting.

“Too many organizations are just ‘paving the cow path’ with AI, applying a thin layer of automation over broken processes and expecting transformative results,” said Tamilla Triantoro, PhD, professor of business analytics at Quinnipiac University and TDWI contributor.

“If you aren’t prepared to redesign your workflows, govern your data, and upskill your people, you aren’t driving transformation - you’re accelerating inefficiency.”

The findings emphasise that organisations must address data governance, skills development and process redesign before expecting measurable returns on AI investments. This aligns with regulatory frameworks emerging globally around AI governance and data management.

The research brief is available at www.tdwi.org. The survey was conducted in June and July 2025 with 155 qualifying responses from various company sizes and industries.