Companies more than doubled AI training. 56% of staff still get no work hours to learn it

Workera published its 2026 State of Skills Intelligence Report on 23 September. The company surveyed 1,000 salaried employees at US organisations with 5,000 or more staff in July 2026 and compared the results with an 800-person survey from March 2025. The share saying their employer offers AI-specific skills training rose from 25% to 58%. The share using AI tools several times a week rose from 40% to 68%. Then the catch: 56% named the same obstacle to building AI skills, no time allocated during work hours. The training budget arrived. The hours to use it did not.

The course teaches the job you already have

The consensus reading of a training rollout is that the employer is handling the transition. Workera's numbers say the employer has bought the content and left the rest to you. In the same survey, 84% of respondents spend five hours or less a week on skill development, and 43% cite a lack of relevant training materials. Nearly 60% say AI skills are weighted equally or not prioritised when their organisation chooses job assignments or promotions. The course exists. The time and the reward mostly do not.

The Federal Reserve Bank of New York shows what that training is for. In a Liberty Street Economics post on 1 September, economists Jaison Abel, Richard Deitz, Natalia Emanuel and Nick Montalbano reported that 61% of service firms in the New York region now use AI, up from 40% last year. Among manufacturers the figure is 51%, up from 26%. Retraining is the main workforce response. Just over a third of service firms using AI, and more than 20% of manufacturers, report retraining staff. Only 4% of service firms laid people off because of AI in the past six months.

Read the content of that retraining. The New York Fed found most firms are helping employees do their current jobs more effectively rather than preparing them for new roles: AI literacy, tool instruction, automating routine tasks, prompt writing. Useful. It makes you faster at the role you hold. It does not tell anyone you are ready for the next one.

Adoption is broad across firms and narrow inside them

The New York Fed figure the headlines skipped is the median share of workers who actually use AI at firms that have adopted it: 17% at service firms, 7% at manufacturers. Adoption is wide at the level of the company and thin at the level of the desk. The same survey found 13% of service firms hired more workers because of AI, while 15% hired fewer. The movement runs towards the minority already doing the work with it.

Workera's data shows why that minority is hard to spot from above. 43% of respondents say they have been passed over for an opportunity or promotion because someone misjudged their skills. Only 27% get regular coaching from a manager or executive, and 22% say they never receive individual coaching. When skill is invisible and coaching is thin, a completed module reads as attendance. A before-and-after record of changed output reads as capability.

Taylor Sullivan, Workera's VP of Product and Assessment, put the responsibility where it belongs: "The fix belongs to the employer." On principle she is right. An operator plans for the employer not fixing it this quarter.

AI does the routine work. You do the thinking. The company course covers the first half of that sentence. Nobody has scheduled time for the second half, and the second half is what gets someone chosen for the next assignment.

Your Next Move

1. Put the hours in writing this month. Ask your manager for a fixed weekly block tied to one named workflow and one deliverable, and put it in your calendar as work. A request attached to an output is easier to approve than "learning time", and it turns development from an extra into part of the role.

2. Train on live work. Pick one recurring task and run it with AI for four weeks. Log the time taken, the errors you caught and the calls you made that the tool could not. The New York Fed found firms stressing verification of AI outputs and avoiding over-reliance. Make checking and judgement the skill people see you exercise.

3. Keep your own skills record. Workera found 41% of employees would opt into continuous skill measurement if they controlled the data. Do not wait for the platform. Keep a one-page log of what AI changed in your output, and bring it to your next review and every assignment conversation. If your organisation does not yet weigh AI skill in promotions, attach yours to results it already counts.