Most companies say they have been blindsided by unexpectedly high AI bills

Asana data reveals that 82% of companies have faced unexpected AI bills in the last year
Risky AI outcomes have affected more than half (53%)
AI systems lack context and workers are forced to pick up the pieces

Four in five (82%) UK IT leaders say their organization has experienced unexpected or unplanned cost increases related to AI in the last 12 months, with many having invested heavily in the technology but unable to predict the true extent of costs, including deployment, governance, integration and scaling.

The data, revealed by Asana, suggests that companies are indeed moving from pilot adoption to scaled adoption, but are not considering the cost changes associated with a broader rollout.

Organizations are also struggling to demonstrate their return on investment (ROI), but IT leaders are under increasing pressure to deliver to the board.

AI bills are skyrocketing

“The challenge now is to turn that into measurable business value,” said Christina Francis, head of Asana UKI and Northern Europe.

Three in five (61%) British ITDMs say they are fully or fully responsible for AI-powered business outcomes, but those outcomes don’t always turn out to be so positive. Half (53%) say an AI tool or agent has taken actions in the past year that have gotten them into trouble, such as financial harm, legal trouble, or reputational damage.

Actual implementations may not be as well thought out either, because half (46%) say AI projects often fail because they lack full context without access to workflows, internal knowledge, and business processes. More than a third (37%) of knowledge workers even spend more than 30 minutes each day correcting AI results due to a lack of context.

“AI is most powerful when it has context: the goals, decisions, and workflows surrounding the work,” Francis added.

Asana also discovered the extent of shadow (or unsanctioned) AI, which makes companies pay for services that are not always used. One in four (25%) say they frequently use unauthorized AI, and two-fifths (38%) regularly use personal AI accounts for work-related tasks.

Ultimately, the report concludes that workers are willing to accept faster ways of working, but will also find alternatives if the approved tools are not up to the task. As always, the best advice is for companies to meet workers where they are and build a strategy around them to deliver the best return on investment and cost savings.

Google logo on black background next to text that says