Cappfinity examines how AI is being used in each stage of talent acquisition today, and how it might be used in the future to inform end-to-end recruitment strategy development.
Artificial Intelligence (AI) has burst onto the scene of talent acquisition. With all the excited noise, it is understandable that people get swept up in the AI hype. But what is really happening, how is AI actually being used, and what do we need to take into account for talent acquisition to leverage this new AI-enabled world?
These are the questions we consider in our embrace of AI, as we explore and figure out when and how to use it, what we need to watch out for, and what the future might hold.
What do we mean by AI?
A good place to start is to consider what people mean by ‘AI’. In simple technical terms, AI is said to exist when the machine can learn from its inputs and evolve its outputs toward an overall goal, refining the actions that it takes over time, based on this feedback loop.
This is the essence of ‘machine learning’ on which artificial intelligence is built. If we were to be strict, we could say that if the machine doesn’t learn, then it’s not AI – but it still could be automation or algorithmic processing.
The difference with automation and algorithmic processing is simply that the machine repeatedly does what it has been programmed to do. It does not, as in the machine learning case, evolve how it does it, from its own learning, in pursuit of a higher order goal (this would be AI).
If you hear the words machine learning, learning model, large language model, generative AI, chatbots or agents, there is a very good chance that these are AI in a technical sense.
If you hear automation or algorithms, there could be AI, but it could also be simply rote-automated programming.
This distinction matters because there are different legal considerations and consequences when machines are making the decisions (AI) rather than humans (programmed automation or algorithms).
When we consider how AI is being used in talent acquisition today, and how it might be used in the future, candidate experience is key.
As a general principle, our experience has shown that there are often careful trade-offs to be struck when using AI to improve efficiency but without damaging the candidate experience of completing the selection journey.
Attraction
Today attraction is typically employer-driven (let’s promote the employer) rather than individual-driven (let’s attract the specific individual). There is also something of a middle-ground, with specific cohort-focused attraction (especially for underrepresented groups).
In the future, attraction will shift to be far more skills-specific, identifying those candidates who have the skills you need, and tailoring your attraction messages to them individually and specifically.
This could be done through things like avatar-driven interview experiences, using immersive VR-driven simulations of the organisation and its opportunities, to deliver the ‘lived experience’ of really ‘feeling like you are there’.
This is the principle that all emotion-based advertising works from (think holidays, entertainment, sports) and it will become a differentiator for employment attraction.
Preparation and matching
This currently includes the use of AI in things like practice assessments to give people exposure to what they are likely to encounter within the selection process, as well as matching tools to help candidates align their interests, values and motivations with the employer overall, or specific roles or pathways within that employer’s offer.
In the future, preparation and matching are likely to include AI-prompted improvements to practice interview questions, avatar-assisted career coaching, and talent pooling skills matches across an employer overall, as well as wider supplier and partner ecosystems.
This talent pooling will be enabled by deeper skills identification and proficiency assessment, mapped to skills models and frameworks that have been defined across the different requirements of the organisation.
This will help employers be far more efficient and effective in their deployment of talent. It will also help candidates to be far more confident in finding the right role and match for their skills. As long as candidates trust the AI-driven matching capabilities of the talent pooling, they can be their authentic self in the knowledge that the machine will do its work and the organisation will have opportunities that are a match for them.
Applications
Today applications are often processed with rule-based screening and points generation systems that determine whether a candidate should progress or not. This can include so called ‘killer’ or ‘critical’ questions, that act as a binary ‘in or out’ for moving a candidate forward.
In the current state, these application systems can often be gamed by knowing ‘what to say’, and this is one of the key battles of ‘candidate AI’ versus ‘employer AI’ – and one of the key reasons for the increase in application volumes and the attendant decrease in quality.
This is a situation that serves nobody’s interests, and wastes everyone’s time (but made a lot easier by having the AI waste its own cycle time).
In the future, AI will be used to drive advances in candidate verification that overcome these problems, filtering out the chaff of inappropriate applications far more effectively.
This candidate verification is likely to include steps that verify human input rather than machine output (think the next evolution of Captchas), steps that verify the claims of the human (for education, skills or experience), and steps that deliberately introduce some friction into the process – easy for a human to navigate, far more difficult for an AI to overcome.
Online assessment
This has delivered massive efficiency, quality, consistency and cost gains to employers for almost two decades now. This is the domain of automation and algorithmic processing.
Online assessment has allowed employers to broaden their application pipelines and so genuinely level the playing field – candidates can apply for anything from anywhere, rather than being pre-filtered based on the university they happened to attend or the degree class they were able to achieve.
In the future, online assessment will support enhanced self-led assessment navigation for individualised and personalised assessment experiences. The candidate’s pathway through an assessment will evolve and vary according to their skills proficiency and propensity, sorting and filtering the roles open to them as they are matched in real time to the opportunities where they are a fit.
Another area where AI is playing a critical role in online assessment (in all its forms) is to counter the inappropriate use of AI in assessment, job simulation or interview responses.
Furthermore, AI is often used in data analytics to identify variations from normal distributions of behaviour to highlight candidates of concern – very similar to how fraud analytics work currently in financial services.
For example, the Cappfinity Candidate Integrity Tracker reports monthly on variations in candidate behaviour and assessment scores. Notwithstanding some claims that ‘candidates are using AI across the application process’, this is not what we see in the data. Across more than 2.5m assessment completions, there is no evidence for consistent score increases for online assessments.
Video interviews
Asynchronous video interviews in particular, where the candidate and interviewer can interact at different times and in different places, have been a great leveller for candidate access, as well as delivering cost, efficiency and consistency gains.
Video interviews today can use AI in the AI-enabled generation of new interview questions, or the AI-enabled scoring of video responses (albeit with some legal controversy and concern around face recognition and emotion detection sensitivity, with particular risk of impacts on underrepresented groups).
In the future, video interviews will become a more seamless interactive interview experience, with avatars taking on the role of the interviewer, and interview questions, content and flow evolving in a uniquely personalised way for each candidate, as the system learns and evolves based on the candidate’s previous responses.
These evolutions of questions, direction and trend in the interview will be happening in real time as the interview is being conducted, mapping the candidate’s skills, experiences, motivations and interests to the plethora of roles and frameworks programmed into the background.
This will act as a sort of a magical sorting hat to guide the candidate to where they are best matched.
Task and job simulations
Traditionally, these might have included analyses, or calculations, or the development of strategies or marketing plans – all micro examples of the types of activities that realistically could be required in the role.
As task and job simulations have evolved, we are now designing them to include the explicit use of AI tools and assistants as a natural and expected part of the role.
These AI-enabled job simulations invite candidates to use their AI assistant of choice to support them in completing the task that has been set. They are then invited to reflect on their process of using these tools, looking at how they critiqued the output, refined their prompt engineering inputs, and ultimately took responsibility for what was delivered as an agentic human being.
In the future, task and job simulations are likely to evolve based on a candidate’s previous responses and submissions, as the AI learns from what has been shared so far and adapts towards the assessment goal that it is designed to optimise for.
In this case, it will be an ongoing challenge to determine where the AI ends and the human begins – a microcosm of the AI-human agentic dividing line that will be forever evolving.
Onboarding and development
Avatar-guided pre-boarding and induction activities support new joiners with what they need to know ahead of starting and in their first days and weeks.
This includes online toolkits and development interventions that are designed to build on and enhance the skills assessed as part of the selection process, and to set people up for faster time to competence and the confidence for effective contribution once in role.
In the future, onboarding and development will take much greater advantage of advances in the availability of Virtual Reality simulated worlds and experiences. New joiners will experience global office tours, meet their future colleagues and hear a welcome message and strategic priorities from their boss, all without needing to leave their own home.
As bandwidth and technological capabilities permit, these will move from being asynchronous, determined events to live, synchronous, free-flowing experiences, with multiple participants all interacting with each other and at play from their remote locations anywhere around the world.
Keeping abreast of AI developments, being aware of the applications but also the limitations of AI, and remaining cognizant of the legal, ethical and practical considerations of AI use, are incumbent on all of us.
Ultimately, the key responsibility we should take away is that of our ultimate, individual, human agency. We can use AI to assist us, but the decisions are ours, for which we bear ultimate responsibility – just ask Steven Schwartz.
5 tips for embracing AI to accelerate recruiter outcomes and candidate experience
- Share with candidates ideas for the best prompts to explain or prepare them for every stage of the assessment process.
- Explain the pros of cons of using AI during the recruitment process and be explicit when you don’t want a candidate to use AI in the process, as well as when they can.
- Be transparent with candidates when you are using AI in your process, for any part of the scoring or capturing of information. (This is often a GDPR requirement anyway).
- Monitor where the use of AI might negatively impact on candidates’ performance and ability to be themselves (e.g. reading from a script during a video interview). Look for ways to encourage preparation and authenticity of responses.
- Use AI and augmentation to develop a more personalised and efficient candidate experience, for example, reducing the number of stages to be completed and so shortening the time candidates spend applying.
This content is part of a Cappfinity whitepaper developed for ISE members, Embracing AI in Talent Acquisition – Today and in the Future
0 Comments