A job can change before its job description does. If AI takes on part of the work, what should you ask of the person doing it? And how will they learn what comes next?
Those questions came up in several Beyond 2026 roundtable discussions. Leaders debated how to define work, hire for it, and help people grow as it changes.
The ideas ranged from teams built around different strengths to hiring tasks that allow AI. Some were still taking shape. Others met pushback, including whether hiring teams could keep up with more fraud checks.
Here are five lessons to help you review your next role, hiring process, or learning plan:
- Define the work first, then look for skills your team lacks.
- Ask candidates to explain their work and tool use.
- Weigh each hiring check against its cost to your team and candidates.
- Give people time to try AI on real tasks, then check if it helped.
- Keep hands-on practice in roles and career paths as AI takes on more tasks.
Which work needs doing, and who should own it?
An open role puts you under pressure to hire. But before you reuse the old job description, check what work the team needs now.
In one planning exercise, leaders had to choose where to start a skills-based approach to work. Should they map the skills they have or plan for the work ahead? They found they were using workforce planning in different ways. Both paths called for knowing the team’s skills, but a list of today’s skills can’t tell you all your future needs. For your next hire, name the work you need done now and the skills it calls for.
Another discussion questioned whether one person needs every skill on the list. Could people with different strengths do the work as a team?
As one participant put it: “What if we were more agile and we took it step by step and say, “This is the work that needs to be done.””
That idea gives you another way to think about a vacancy. You could look for a skill the team lacks, rather than keep adding to one person’s job. Leaders proposed sharing work this way, but still had to work out how people would know what to deliver when their next assignment wasn’t clear. If you share work this way, set clear tasks for each person. Then decide what a new hire needs to show.
Key takeaway: Start with the work you need done, then look for skills your team lacks.
What should a candidate show you?
A new job description won’t help much if the interview still asks the same old questions. When you open a role, agree with the hiring manager on the work a new hire should deliver and how you’ll judge it.
Leaders shared several ways to make that clearer:
- Set the bar before choosing a candidate. One person shared how they agree on success measures when a role opens. They do it before a favorite emerges, so they can check a later gut-feel choice.
- Give each interviewer a focus. Another person gives each panel member a set of skills to assess. Each can then bring evidence on those skills to the hiring decision.
- Ask how and why. In a separate discussion, one person shared how they let candidates use AI. They ask them to explain their work, then use follow-up and “what if” questions to probe how well they understand it.
- Make tool use clear. Another participant had planned a project exercise that allowed AI: “The expectation is you share what you did, what tools you used, and why.”
The finished work gives you a place to start. Asking someone to explain it helps you explore what they understand and what they did themselves. Then comes the next question: how do you check what they’re telling you?
Key takeaway: Set clear goals for the role, then ask people to explain their work and tool use.
How do you check a candidate’s work without asking too much?
How do you know the person who starts a job is the person you interviewed? In a roundtable exercise, leaders weighed ways to spot an identity swap against the time and cost of more checks. That is a different problem from judging how well someone used AI.
Participants discussed checking details by hand. But who has the time and skills to do that well? One person called for more training and support for those who judge candidates. Another questioned how practical it would be to train everyone, from the first screen to the final interview. The sticking point was how to support all those people. Before you add a check, decide who can review it well and what help they’ll need.
The burden on candidates matters too. An in-person step could keep out people who can’t afford to take part. A work trial came up as a secondhand idea, along with this question: “You can’t just have them do the job for free, right?” A trial might give you more to judge, but you also need to consider pay and whether people can take part.
For the identity-swap exercise, people proposed first telling the person what seemed wrong. They would hear the person’s side before deciding what to do. For your team, set out who will check a concern and how you’ll review it.
You may want stronger checks, but each one needs a clear purpose and someone who can carry it out. The training debate shows why adding a step on paper is only part of the job.
Key takeaway: For each check, ask what you’ll learn, who will do it, and who will bear the cost.
How do AI experiments become useful work?
If you ask candidates to explain how they use AI, bring that same focus to the people you already employ. What do they need to use it well in their own work?
One person described starting with team sessions where colleagues showed what they were trying. They gathered feedback, revised selected tools, and shared them with the wider team. People were using them, but the rollout was still in progress. If your team’s trials stay separate, a shared session could help you spot a task worth working on together.
In a separate discussion, someone described setting a shared expectation for AI use, with examples suited to each kind of work. People also had time set aside to learn and try things. If you want employees to learn to use AI, make room for that learning in their jobs.
A tool that works still needs to prove useful. As one participant said: “For me, it’s one thing to build it and deploy it. It’s another thing to actually measure it.”
Did it save time? Did it help the team make better hires? Could the team stop doing part of the work? Use those questions to set a clear goal for your own trial.
Leaders also debated what to do when someone won’t use AI after getting support. One view was that people need to adapt. Another was to reshape the work around the expertise they still bring. If you’re facing that choice, first check why the person isn’t using AI. Do they need more help, or are they refusing to use it?
Key takeaway: Give people time and help to try AI on a task, then check if it helped.
How do people learn enough to take the next step?
Some tasks that AI can take on are also tasks people learn from. If you cut that practice, how will they build the skills to take on harder work?
One person worried that the rush to get more done could crowd out time to learn: “I just wonder if we’re going to be so impatient and so enamored by the productivity increase that we’ll skip it.”
To keep that chance to learn, people suggested safe simulations. These would let learners make mistakes and learn from them. They also discussed having people do some work by hand before handing it to AI. For your learning plan, ask which tasks people still need to try themselves so they can judge AI’s work.
That need to build judgment extends to career growth. As AI takes on more tasks, what should people learn to take on harder work and earn higher pay? One participant had started to tie levels and pay to a person’s contribution, rather than the size of the team they managed. That raised a concern: would higher output goals still leave room for new ideas, mistakes, and working together? If you change how people advance, leave room for the practice that helps them grow.
This brings us back to how you share work across a team. A new task can be a chance to learn as well as a job to finish. Learning in context ties practice to work you need to do. When you review a career path, ask what people can practice along the way.
Key takeaway: Keep time for hands-on learning when you change a role or career path.
Choose a decision to revisit
When AI changes the work, review the choices you make about skills, hiring, and learning too.
Start with one role. With relevant SMEs, name the work that has changed and what the person now needs to deliver. Choose one change to try: update a hiring task to ask how someone reached an answer, or give employees time to practice a skill the role still needs.
Set a date to check whether that change helped. Review the work and ask the people involved what worked or got in their way. Use that evidence to decide what to keep or change.