Only 28% of Trained Workers Keep Using AI, Survey Finds

Brought to you with backing from

A training session does not guarantee that a tool gets used afterwards. In a TalentLMS survey of 1,500 full-time US employees, only 28% said they consistently use AI after being trained on it.

Managers who use the tool themselves

Employees were more than twice as likely to use the new approach consistently when their managers used it too: 49% against 22% where managers did not.

Carolina Caro, CEO of Conscious Leadership Partners, said in the report about what she calls "identity inertia", which she described as "the pull toward what is familiar, even when we've learned something new." Caro said managers can counter it by modelling the new behaviour themselves and making their own moments of reversion visible.

Why people go back

TalentLMS calls the broader pattern "habit hugging": sticking to familiar routines as jobs change and AI reshapes the work. It reaches beyond AI. Across every kind of new approach, 41% of employees use it consistently after training, 52% follow it but sometimes fall back, and 8% do not adopt it.

Dropping an old habit is more difficult than picking up a new one for 65% of respondents. Some 42% still use tools or processes that their manager thinks have been replaced. People who kept or returned to old methods said the old way feels more familiar (55%) and is faster (48%).

Pressure plays a part. Six in 10 said results count for more than the new method at their company. Another 48% feel rushed to adopt new ways, and 41% have room to be less productive while learning.

Where AI stands

All respondents had received training on a new tool, technology, process or way of working within the previous 12 months. For AI, 36% use it but sometimes revert, and 12% were trained yet still work without it. A further 24% said they were never trained to use AI for work tasks.

Lack of trust in AI for high-stakes tasks was the top reason for not using it consistently, at 32%. Spending too long checking or reworking the output came next at 25%. Fifteen per cent were unsure when AI use was acceptable, and 15% forgot.

What employers follow up

Companies track completion more often than use. About 41% of employees said their employer records whether training was finished. Fewer reported checks on whether they apply it (36%) or understood it (34%), and 10% said none of the listed follow-up activities happen.

Asked what would help, half pointed to making the new way part of everyday workflow. Forty-nine per cent wanted more chances to practise, and 47% wanted to see how it improves their work. Dimitris Tsingos, CEO of TalentLMS parent Epignosis, said "AI training alone isn't enough to make new ways of working stick."

Our View

These are self-reported habits, not measured output, so they show perception rather than productivity. Still, the gap between counting completions and checking use stands out. The survey does not separate real caution, such as the 25% who spend time checking AI output, from plain habit. Whether that checking is waste or sensible control depends on the task, and each team's managers would have to decide.

Source: Human Resources Director