If you wait until implementation, some of the most important decisions may already have been made. When organisations introduce AI at work, involving employees seems like an obvious part of doing it responsibly. Explain what the technology will do. Talk to employee representatives. Provide training. Listen to concerns. Help people adapt to new ways of working.
But what if that's already too late? That was one of the questions raised at Arendalsuka, Norway's annual gathering for public debate and discussion on issues shaping business and society. During the session “Hvem bestemmer når KI brukes på jobb – ledelsen, de ansatte eller teknologien selv?”, organised by Delta, the discussion explored what happens to influence and decision-making when AI becomes part of HR, management and how work is organised. One idea in particular stood out:
If employees only get a say when AI is being implemented, some of the decisions that will affect their working lives may already have been made.
The important decisions can happen before the AI is switched on
When we think about AI making decisions at work, we tend to focus on the output. Which candidate did the system recommend? How did it create the shift plan? Why did it allocate that task to this employee? But Lotte Skolem, Head of Emerging Tech at Langsikt, raised an important point during the discussion: influence can move much earlier in the process. It starts when an organisation decides which system to buy.
What should the system do? Which data should it use? Which parameters should matter? How will it be tested? Where will it be allowed to influence decisions? These might sound like technology or procurement questions. But many of them are also people questions. And once the technology has been selected and configured, some of those choices may be much harder to change.
What does a “good” outcome actually mean?
An example shared by Trond Ellefsen, President of Delta, made this particularly tangible. AI-supported shift planning can bring real benefits. It can reduce administration, make planning faster and help organisations use limited resources more effectively. But imagine asking a system to create the most efficient possible shift plan. What does “efficient” mean? Perhaps it means covering every shift while making the best possible use of available resources. But there are other considerations.
Which colleagues work well together? Where is particular expertise needed? Is there enough continuity? How does the schedule affect the working environment? Are there individual circumstances that should be considered? As Ellefsen pointed out during the session, a system can optimise staffing without necessarily optimising the working environment or the quality of the service being delivered. And that leads to a bigger question:
Who gets to decide what the AI should care about?
AI doesn't decide what matters on its own. People make choices about what the technology should prioritise – either within the organisation or through choices already built into the system. And those choices can have a very real impact on people's working lives.
There's another voice at the table
The session also raised an aspect of AI at work that receives less attention. Many workplace decisions involve a relationship between employer and employee, with employee representatives playing an important role in decisions that affect working life. But AI introduces another important actor into that relationship: the technology provider. This was discussed at Arendalsuka in the context of Norway's long tradition of collaboration between employers and employees. But the underlying question is much broader: what happens when some of the choices that shape people's working lives are embedded in technology supplied by a third party?
Skolem raised the possibility that influence could move first from employees towards employers, and then from employers towards technology providers. Ellefsen made a related point. As decisions about areas such as scheduling, staffing, recruitment, competence development and work processes become embedded in digital systems, some of the assumptions shaping work may sit with a technology company that has never met the employees using the system. That doesn't mean organisations are handing every decision to their software provider. But it does mean HR should understand what has already been decided by the technology – and what the organisation can decide for itself.
For example:
What assumptions does the system make?
What can we configure ourselves?
What can't we change?
What is the system designed to optimise?
These aren't only questions for IT.
Employee involvement may need to start earlier
Kjartan Almenning, Director of Industrial Policy at Abelia, made the timing issue particularly clear during the discussion. His argument was that if employee representatives first enter the conversation when the technology contract is ready to be signed, it isn't meaningful participation in the decision. At that point, it is closer to being informed about a decision that has already been made. The exact mechanisms for employee involvement differ between countries and organisations. But the underlying principle is relevant much more widely: involvement only has real influence if it happens while there are still decisions to influence.
The alternative is to involve people earlier. Not because every employee needs to participate in selecting technology. But because the people who understand the work often understand things that aren't visible in a system specification. Anders Karde, Head of Data and Digital Engagement at Nordea, described how some of Nordea's strongest results from AI have come from working closely with employees who understand the problems the technology is intended to solve. That collaboration can help organisations identify where automation makes sense, where AI can support employees and where better access to information could improve their work. Employee involvement then becomes more than something that happens after a technology decision. It can become part of making better decisions about the technology in the first place.
This is a question for HR – before implementation
HR is often heavily involved when new workplace technology is rolled out. There is communication to manage. Employees need training. Roles may change. Managers need support. Questions and concerns need to be addressed. With AI, HR may need to move further upstream. Before selecting a system that will influence employees or decisions about them, there are some basic questions worth asking:
What decisions will this technology influence?
What is it being asked to optimise?
What employee data will it use?
Which decisions will remain with managers?
Can employees understand and challenge recommendations that affect them?
Have employees or their representatives had a meaningful opportunity to influence how the technology will be used?
These questions aren't about slowing AI adoption down. In fact, the discussion at Arendalsuka repeatedly returned to the opposite argument: organisations need to make use of AI, and the tradition of collaboration between employers and employees could be an advantage in doing that well.
Getting involved before the decisions are made
Perhaps the most useful learning from the Arendalsuka discussion is that employee involvement in AI isn't simply a question of whether people are included. It's also about when. If employees are brought into the conversation once a system has been selected, configured and is ready to launch, they can still provide valuable input into implementation. But they may have missed some of the decisions that matter most.
- What should the technology do?
- What should it optimise?
- Which decisions should it influence?
- And what should remain a human judgement?
As AI becomes embedded in more parts of working life, these questions will increasingly shape people's everyday experience of work. If we want employees to have a meaningful say in that future, the conversation needs to start before the technology arrives.
About this article
This article is part of our The human side of AI series ,inspired by sessions at Arendalsuke, Norways annual gathering for public debate and discussion on issues shaping business and society. It draws on the session “Hvem bestemmer når KI brukes på jobb – ledelsen, de ansatte eller teknologien selv?”, with Lotte Skolem, Head of Emerging Tech at Langsikt; Kjartan Almenning, Director of Industrial Policy at Abelia; Trond Ellefsen, President of Delta; and Anders Karde, Head of Data and Digital Engagement at Nordea. The session was moderated by Hanne N. Skjønsberg, Head of the Working Life Department at Negotia/Delta.
Source: Arendalsuka – Hvem bestemmer når KI brukes på jobb – ledelsen, de ansatte eller teknologien selv?
