The ABP is hosting a series of sessions this year, addressing Generative AI and the Future of Business Psychology, as discussed in the article: Why Generative AI Matters to Business Psychology. The series organiser, Dexter Winters, is the AI Practice Insights Lead for the Association for Business Psychology (ABP).

In each session, experts address an Area of Practice of Business Psychology, as outlined in The ABP's Unified Framework. This article provides an overview of the session on the first Area of Practice:  Selection & Assessment.

AI and Assessment & Selection

What Is Changing, What Matters, and What Comes Next

Artificial intelligence is already reshaping how organisations select and assess people. The first session in the Association for Business Psychology’s series on Generative AI focused on what this means in practice.

The discussion brought together Nigel Guenole, Kate Young, Keran Dhillon, and Luke Montuori. It explored how AI is being used today, where risks are emerging, and how practitioners can respond.
The message was consistent. AI is not a future issue. It is already changing behaviour, decision making, and expectations across the entire selection process.

The current state: rapid change and uneven standards

A clear theme from the discussion was the speed of change. AI has lowered the barrier to creating assessment tools, but not the standard required to use them responsibly. Kate Young described the scale of this shift:

“Someone can layer Google Studio on top of a prompt, on top of a foundational LLM, and call it a hiring tool. That is genuinely the state of play today.” 

This creates a gap between what looks plausible and what is actually valid. Nigel Guenole reinforced this point:

“The gap between being able to imagine something and being able to build something that looks plausible is really, really small now.” 

At the same time, many vendors entering the market do not have psychological expertise. This shifts influence away from evidence-based practice and towards marketing and commercial pressure.

Disruption across the whole selection pipeline

AI is not affecting one stage of hiring. It is reshaping the entire system. Luke Montuori highlighted how behaviour is changing on both sides:

“People aren’t just using AI to optimise their answers, but they’re also using AI to rethink how they approach tasks and work more generally.” 

This applies equally to employers. AI is now used for screening, interpretation, and decision support. The result is a feedback loop where both candidates and organisations are adapting to AI at the same time. One consequence is that traditional assessment approaches are becoming less stable. As Luke noted:

“AI has made static or one-shot assessment processes more brittle.” 

This challenges long-standing assumptions about how and when measurement takes place.

The risk landscape: validity, fairness, and governance

The panel identified several practical risks that are already visible in organisations.

1. Loss of measurement rigour

There is growing use of AI tools without clear evidence of validity or reliability. This includes summarising CVs, interpreting assessments, and analysing interviews. Luke described a common pattern:

“As long as there’s a surface level of plausibility, as long as it’s face-valid, then that’s fine.” 

This creates a false sense of objectivity. Outputs appear credible, but may not measure what they claim.

2. Governance gaps

AI tools are often introduced through multiple functions, including procurement, legal, and HR. Responsibility for assessment quality can become unclear. Keran Dhillon explained the issue:

“There is no one person that says, I need to hold all three of these strings accountable, and then build in that measurement science.” 

This leads to decisions being made without full consideration of psychological standards.

3. Bias and unintended impact

AI introduces new risks alongside existing ones. These include indirect discrimination and impacts on specific groups. Keran highlighted concerns around neurodiversity and communication patterns:

“What’s weighted and what’s not weighted? There is no real clarity… that becomes a measurement validity problem.” 

Even well-intended uses, such as summarising interviews, may introduce bias if not carefully evaluated.

4. Candidate behaviour and system distortion

Candidates are increasingly using AI to support applications. This affects the integrity of assessment data. Nigel observed measurable changes:

“You do see an uptick in the proportion of people getting particular cognitive ability items right.” 

At the same time, AI-generated CVs are becoming harder to differentiate:

“They all come in with these CVs that are perfectly formed… it’s really, really hard.” 

The role of the business psychology practitioner is changing

A consistent message was that business psychologists need to adapt. Keran described the shift clearly:

“We can’t govern what we don’t understand.” 

This means developing new capabilities alongside existing expertise. These include:

• Understanding how AI systems are built and trained
• Interpreting algorithmic outputs and audit results
• Translating technical risk into business language
• Engaging with regulation across jurisdictions

There is also a need to influence practice more directly. Rather than acting only as gatekeepers, psychologists must become enablers of responsible use.

What should remain constant

Despite the disruption, the panel was clear that core principles still apply. Nigel emphasised the value of established methods:

“We’ve been working with those methods for 100 years… we really know what good looks like.” 

Validity, reliability, fairness, and transparency remain central. AI does not replace these standards. It increases the need for them.

Where there is genuine opportunity

The discussion was not only about risk. There are meaningful opportunities if AI is applied well.

1. Better assessment design

AI enables more adaptive and interactive assessments. Luke described the potential:

“Assessment can become more adaptive and conversational… really measure process.” 

This allows richer data and a move towards “systems of evidence” rather than single test scores.

2. Continuous validation

AI systems can support ongoing monitoring rather than one-off validation studies. This could shift practice towards continuous evidence generation and stronger quality control over time.

3. Scalable and personalised assessment

AI can support the creation of structured, skills-based assessments at scale. Keran shared a practical example:

“We’re able to provide more ad hoc, personalised assessments… where previously, that has been very costly.” 

This has clear benefits for both organisations and candidates if governed properly.

4. A stronger role for psychology

AI creates a need for expertise in behaviour, measurement, and ethics. As highlighted in the session framing:

“This is fundamentally a human challenge, not a technical one.”

This positions business psychology as central to responsible AI adoption.

A critical tension: speed versus standards

The central challenge is how to balance innovation with rigour. Nigel summarised this tension:

“Psychologists need to speed up a bit, and technologists need to slow down a little bit.” 

This “messy middle” is where most organisations now operate.

Final reflections

AI is not simply another tool in selection and assessment. It is changing how people think, behave, and make decisions.

The key risks are already visible. So are the opportunities.

For business psychologists, the implication is clear. Engagement is not optional. The profession must actively shape how AI is used, grounded in evidence and aligned with established standards.
Or, as the session made clear, if psychologists do not take that role, others will.