Your next boss might be software. Should you worry?

What does an AI boss actually do?

Usually, it handles a slice of management rather than the whole job. Recruitment software can sort CVs or score answers to screening questions. Scheduling systems can allocate shifts. Monitoring tools can collect information about attendance, activity or completed work and present it as a performance score.

Some of this is useful. A consistent first pass through a large pile of applications may save recruiters time. A rota that accounts for availability could spare everyone the Sunday-night spreadsheet ritual. If a system spots a missed delivery or an overloaded team early, a decent manager has a chance to put things right.

But an algorithm — a set of rules for turning information into an output — is not necessarily artificial intelligence. And calling a tool ‘AI’ tells you remarkably little about whether its judgement is sound. The important questions are what it measures, what it misses and what the employer does with the result.

There is a substantial difference between software suggesting that a manager look at an application and software rejecting it automatically. In its 2026 examination of recruitment practices, the Information Commissioner’s Office (ICO) found that some employers using automation were likely making significant decisions without meaningful human involvement. The regulator also found shortcomings in how candidates were told about those decisions and how they could challenge them.

Can an algorithm be a fairer manager?

Potentially, in a narrow sense. A well-designed system might apply the same relevant criteria to every application, rather than favouring someone because their CV reminds an interviewer of his own. Speed and consistency are real advantages. They are not the same thing as fairness.

A hiring tool trained on past choices can reproduce past preferences. Even a simple rule can cause trouble: treating a gap in employment as a mark against someone may overlook caring responsibilities, illness or a career change. If you have spent decades doing the work but describe your skills differently from the job advert, a rigid screen may miss you altogether.

A computer can make a bad decision consistently. That does not make it objective.

Performance scores have a similar blind spot. Counting calls is easy; judging whether you patiently solved the difficult one is harder. A dashboard may show that a colleague completed fewer tasks without showing that they trained a new starter or dealt with the customer nobody else could help. If managers start treating a convenient number as the whole job, the tool has stopped assisting them and started narrowing their judgement.

There is also a practical test for employers: does a human reviewer have the time, information and authority to disagree? If the answer is no, a person clicking ‘approve’ at the end is not much of a safeguard. It is management by autopilot, with someone available to take the blame.

What are your rights if software makes the decision?

Across the UK, data protection law matters when an organisation uses your personal information to make decisions. From 5 February 2026, changes brought into force under the Data (Use and Access) Act 2025 widened the circumstances in which organisations can make significant decisions using solely automated processing. They did not give employers permission to do it without safeguards.

If a decision based solely on automated processing has a legal or similarly significant effect on you — a consequential hiring decision, for example — the UK GDPR requires safeguards. These include information about the decision and opportunities to put forward your view, obtain human intervention and contest it. Whether those specific rules apply depends on what the system actually did and whether a human played a meaningful part in the decision. An AI-assisted decision is not automatically a solely automated one.

You can ask an employer or prospective employer what personal information it holds about you through a subject access request. That can help you understand the information used in a process, although it is not a magic key to every line of a supplier’s software. Ask separately whether an automated tool made or supported the decision, what information it used and who can genuinely review the outcome.

Discrimination law applies too. In England, Scotland and Wales, age and disability are protected characteristics under the Equality Act 2010. Northern Ireland has separate equality law: its Employment Equality (Age) Regulations (Northern Ireland) 2006 protect people against age discrimination in recruitment and employment. Putting software between you and an employer does not make unfair treatment acceptable.

If an application system creates a barrier because of a disability, ask for an adjustment rather than struggling through it in silence. Government guidance on AI recruitment warns employers that automated screening and digital assessments can disadvantage applicants, and that reasonable adjustments may sometimes mean offering another way to take part.

Keep the manager in the room

If you are an employee, ask what the system measures before a score becomes part of an appraisal. Does it count quality as well as quantity? Can it distinguish approved leave from absence, or complex work from slow work? Who checks errors, and how often? Those are ordinary management questions, not a declaration of war on technology.

Keep your own record of achievements that a dashboard cannot see: difficult problems solved, colleagues supported, work improved. If a score looks wrong, challenge the inputs and the conclusion separately. ‘The system says so’ is an explanation of where a number came from, not why it deserves to govern your career.

If talking it through does not resolve the problem, check your employer’s grievance procedure and put the concern in writing. Acas advises employees in England, Scotland and Wales to set out the issue, any evidence and what they want the employer to do. In Northern Ireland, seek advice through the relevant local employment and equality bodies rather than assuming Acas procedures apply.

For managers, the sensible bargain is simple: let software do the repetitive sorting, then spend the time saved on the human parts of the job. Listen, investigate odd results and take responsibility for decisions. If you cannot explain a verdict without pointing at a screen, you are not managing your team. You are operating its printer.

In short

Software can make parts of management quicker and more consistent, but it is not a reliably fair replacement for a good boss. Treat its scores as evidence to examine, not verdicts to obey. If an automated decision significantly affects your job prospects or working life, ask how it was made and how a person can review it.

Sources

• Information Commissioner’s Office – 2026 report on automated recruitment and safeguards
• GOV.UK – guidance on responsible AI in recruitment
• legislation.gov.uk – commencement regulations for the 2025 data law
• Equality Commission for Northern Ireland – guidance on age discrimination at work
• Acas – guidance on raising a formal grievance