Most AI Project "Failures" Are Fixable Growing Pains, Not Dead Ends, study finds

Info-Tech Research Group has published a new blueprint, "Get and Keep Your AI Projects on Track," aimed at helping IT leaders evaluate struggling AI initiatives and decide whether to continue, redirect or stop them, the firm said. The research comes as rising reports of AI project failure intensify scrutiny around AI investments, making that continue-or-stop decision increasingly difficult for organizations to make.

According to Info-Tech, AI projects face a more intense set of challenges than traditional IT initiatives. The firm said organizations often interpret project setbacks as signs of outright failure, when many of the obstacles AI initiatives face are predictable challenges tied to rapidly evolving technology, shifting expectations, governance requirements and adoption barriers.

"Many of the obstacles organizations are encountering today are temporary growing pains rather than permanent barriers to AI success," said Jenn Aswald, research analyst at Info-Tech Research Group. "IT leaders need to understand which issues require intervention and which will diminish as AI capabilities, governance practices, and organizational experience mature."

The blueprint organizes the most common causes of AI project failure into five root-cause categories, according to Info-Tech. Rapid obsolescence can erode project relevance, the firm said, since AI capabilities evolve quickly enough that solutions risk becoming outdated before implementation is even complete. Value gaps limit business outcomes when organizations struggle to translate AI capabilities into measurable value and sustained operational impact, Info-Tech said.

Completion challenges also delay progress, according to the firm, when AI projects lack structured start, scale and closure criteria and remain stuck in prolonged uncertainty. Conflicting stakeholder perspectives, which Info-Tech calls "fierce opinions," can complicate decision-making by slowing sponsorship alignment and hindering effective governance. And adoption resistance can reduce value realization, the firm said, as previous low-value AI experiences undermine trust and limit engagement with new initiatives.