AI Upskilling Needs Time to Practice and a Way to Prove It

AI upskilling needs room in the working week. If employees finish a course and return to the same workload, with no time to try a different approach, leaders have left the hardest part of learning to chance.

We think the next leadership question is simple. What can someone do better after the training, and have we given them a fair opportunity to demonstrate it?

New research from edX Enterprise and NewtonX gives that question some urgency. It also gives leaders a reason to examine what happens between an AI learning assignment and the next performance conversation.

The gap between priority and capability

In its October 8 research announcement, edX reported that:

• 83% of respondents consider AI capability an active workforce priority.

• 29% say their organizations have mature AI capabilities.

• 78% lack a strong mandate for AI upskilling.

• 26% are highly confident their learning and development function can deliver the capabilities their workforce needs.

NewtonX surveyed 507 verified learning and workforce strategy leaders at organizations with at least 500 employees. The research was fielded globally in June and July 2026, with 10 additional expert interviews.

These self-reported findings come from vendor-sponsored research. The sample doesn't represent smaller businesses, and the findings don't establish cause and effect.

The useful management question is whether people have the conditions to learn. edX describes a strong mandate as executive support, clear capability priorities, and sufficient budget and capacity. Leaders can examine those conditions without treating the survey as a prescription.

Put practice on the calendar and take something off

Protected practice means time during the working week to apply a skill, review the result, and try again. The manager makes that time available by adjusting other expectations.

That last part matters. An invitation to experiment can sound reasonable until someone looks at the deadlines already on their calendar.

For an initial team exercise, you might reserve an hour each week for four weeks. Decide which meeting, report, or lower-priority assignment will move to make room. The right amount of time will depend on the task and the team's workload.

Ask employees where they need help before choosing the exercise. A manager might assume that drafting is the problem when the team spends most of its time checking incomplete inputs.

Include managers in the practice. They need to understand what a useful result looks like and where an employee may need support. Someone has to review the work, answer questions, and keep the learning time from disappearing when the week gets busy.

Describe the capability in terms of the job

“Complete the AI course” is a clear administrative instruction. To assess capability, add a description of the work someone should be able to perform afterward.

Consider a hypothetical account team learning to prepare customer meeting briefs. A useful expectation could be to produce an accurate brief from approved account materials, identify missing information, and explain which statements need checking.

The employee should be able to:

• Select the relevant source material and use it in an approved tool.

• Check the draft against the underlying records.

• Catch an unsupported claim and correct it.

• Explain when the available information is too weak to support a recommendation.

Practice with one account, then assess the same skills on a different example. That helps a manager see whether the person can apply the approach beyond the exercise they already know.

Course completion still belongs in the record. Pair it with a reviewed work sample and feedback on what needs another attempt. Someone who spots a problem and stops should receive credit for that judgment.

Check who benefits from the time saved

Gallup's October 6 findings add an important employee perspective. Among U.S. workers who had used AI in their jobs, 63% said it helped them work faster, while 46% reported higher-quality work. And 31% said their employer had asked them to take on more responsibilities because of their AI use.

Those are employees' assessments, not independently measured productivity gains. The findings raise a question worth discussing before a learning program starts. What will happen if people become faster?

If every saved hour immediately becomes another assignment, employees may have less room to keep developing their skills. Leaders should decide how some of that capacity will support better work, customer attention, or continued practice.

Ask about review time as well. A faster first draft may still require substantial checking. Include that effort when deciding whether the new approach improves the job.

Run a small capability review over 30 days

We'd start with one team and a task that happens often enough to practice several times. A month can provide an initial review point. It won't establish lasting proficiency in every role.

1. Choose the task with the team. Describe the current difficulty and collect a few representative work samples. Agree on what accuracy, completeness, and useful judgment look like before introducing a new method.

2. Make the time available. Name the manager responsible for protecting practice time. Record what will be reduced or deferred, who can answer questions, and which tools and materials participants can use.

3. Practice and review together. Have employees explain their choices as well as show their output. Keep examples of errors, corrections, and occasions when AI wasn't helpful. Use those examples to improve the next session.

4. Assess a fresh example. Review a new work sample against the same expectations. Include preparation, checking, and rework in the time calculation. Ask employees what they can now do independently and where they still need help.

5. Decide what deserves another month. Continue approaches that show useful progress. Adjust the task, support, or training when results are mixed. If the method adds work without improving the outcome, change it before involving more people.

Keep the review proportionate. A short record of the task, time spent, quality issues, and demonstrated skills is enough to begin. Attendance and tool usage can provide context, but the work samples should carry more weight in a capability decision.

Make learning a leadership responsibility

At HiVoltage, we see workforce development as part of AI leadership. Someone needs to connect the learning goal to workloads, coach the managers, and review the evidence with the team.

Before approving the next round of AI training, ask the manager what will move off the calendar and what work the team should be able to demonstrate afterward. Clear answers give people a better chance to turn learning into something they can use.