AI training is informal

getty

AI is supposed to represent our new “co-workers,” right? Well, it would be nice if organizations took the time to help us get to know our new co-workers, which doesn’t seem to be the case.

While two-thirds of companies in a new survey are either requiring or strongly suggesting that employees get to know AI, only 35% of their employees report they have received AI training of any kind from their employers. At this point, only one in six employees feels fully prepared to use it to its full extent.

This represents a yawning gap between AI adoption and employee confidence to use it, according to the survey of 1,000 U.S. employees released by Study.com.

Instead, employees are learning about AI on their own. A majority, 58%, learned more informally – through self-teaching, trial and error, or through colleagues. About 27% first learned through employer-provided training.

More than one in five, 21%, say that AI is now a core requirement for their jobs. They are split on how well AI is working out for them; 48% report they can complete tasks with little to no editing, while 52% need to rework AI output. In terms of time saved on their jobs, the feeling is lukewarm at best. Twenty-nine percent say AI has no time, and another 34% calculate it at two hours or less.

Close to half, 44%, are now comfortable with calling out wrongful AI output. More than two in five are becoming adept at prompt writing to build the responses they need, and a similar number feel they know when AI is doing what it’s supposed to do. Areas where they are flummoxed include applying AI to role-specific tasks, and getting reliable results as early as possible.

Where Employees Are Comfortable with AI

Notice when AI output is misleading or wrong 44%

Write prompts that give consistent results 43%

Judge whether AI output is accurate and useful 43%

Improve weak output with follow-up prompts 36%

Source: Study.com

Where Employees Have Difficulty with AI

Applying AI to complex or role-specific tasks 20%Getting reliable results without repeated retries 20%Receiving inaccurate or false data/information 20%

Knowing how to begin a task with AI 14%Applying AI to complex or role-specific tasks 14%

Source: Study.com

The majority are learning through trial and error, without structure, benchmarks, or feedback loops, the survey shows. Only 32% have a clear standard of what “good” AI use looks like, while the 35% say they have a “rough idea” and 19% simply do not know at all.

Respondents indicated the kinds of training they would welcome:

• Short and applied. “Modules of 30 to 60 minutes, tied to specific job tasks.”
• Role-specific. “Applied examples drawn from employees’ actual responsibilities.”
• Accuracy practice built in. “Include evaluation and fact-checking components — the skill gap employees report most consistently.”
• Clear performance benchmarks. “Explicit standards for what good output looks like.”