Artificial Intelligence-based interview scoring learns from past interview answers, which makes it hard for it to determine if a candidate is legitimately answering the question if their response includes context or an example rarely seen in training data.
Moreover, AI interview platforms may be susceptible to adversarial inputs where an irrelevant answer may receive a high score. Both scenarios raise fairness concerns and can erode trust in AI job interviews (Madaio et al, 2020).
This is why identifying outliers that differ significantly from the majority of answers and flagging them for manual review become crucial steps toward responsible and fair use of AI interview software. While simple rule-based methods (Reiz and Pongor, 2011) could help filter out some irrelevant answers based on answer length and regular expressions, these methods do not take into account the context and content of the answer and question. Someone may describe a very unique, yet relevant, situation in response to an AI for interviews question, which you wouldn’t want to disregard.
In this study, we introduce an unsupervised, question-aware, multi-context outlier detection model that can help detect anomalous answers contextually and semantically. The unsupervised approach is deemed to be more practical compared to a supervised model that requires a large labeled dataset of outlier answers. It helps bootstrap an outlier detector that can then be enhanced through human feedback.
We tested the outlier model to ascertain how well it is able to correctly identify 177,691 actual hired candidate interview answers from outliers, (e.g., movie reviews, news articles, nonsensical text, and sentences generated using BERT (Vaswani et al, 2017) with random starting words).
Our model outperformed the baseline One-class SVM outlier detector (Li et al, 2003), in detecting outliers from actual interview answers. The performance of our model over the baseline unsupervised model can be explained by both question-aware learning and multi-context learning, which help yield better contextual representations for detecting outlier answers from typical interview answers.
We also conducted a human evaluation on 10,689 interview answers of candidates who were not hired and might have provided outlier answers. Our model predicted 0.16% of the answers as outliers with only 5.9% of them being false positives. All of these false predictions describe contexts related to family and personal life in their answers but are relevant to the question. It is reasonable that these answers are labeled as an outlier by our model since they are contextually and semantically different from most interview answers.
While a data-driven AI interviewer can help counter flaws in human interviewers, answers that are significantly different to training data can lead to spurious predictive outcomes. In this study, we show how a
References:
Dai, Y., Qi, J., & Zhang, R. (2020). Joint recognition of names and publications in academic homepages. In Proceedings of the 13th International Conference on Web Search and Data Mining (pp. 133-141).
Li, K. L., Huang, H. K., Tian, S. F., & Xu, W. (2003). Improving one-class SVM for anomaly detection. In Proceedings of the 2003 international conference on machine learning and cybernetics (IEEE Cat. No. 03EX693) (Vol. 5, pp. 3077-3081). IEEE.
Madaio, M. A., Stark, L., Wortman Vaughan, J., & Wallach, H. (2020). Co-designing checklists to understand organizational challenges and opportunities around fairness in AI. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1-14).
Reiz, B., & Pongor, S. (2011). Psychologically Inspired, Rule-Based Outlier Detection in Noisy Data. In SYNASC (pp. 131-136)
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., … & Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30.
There’s growing interest in AI-driven tools that infer skills from CVs, LinkedIn profiles, and other passive data sources. These systems claim to map someone’s capability based on the words they use, the jobs they’ve held, and patterns derived from millions of similar profiles. In theory, it’s efficient. But when inference becomes the primary basis for hiring or promotion, we need to scrutinise what’s actually being measured and what’s not.
Let’s be clear: the technology isn’t the problem. Modern inference engines use advanced natural language processing, embeddings, and knowledge graphs. The science behind them is genuinely impressive. And when they’re used alongside richer sources of data, such as internal project contributions, validated assessments, or behavioural evidence, they can offer valuable insight for workforce planning and development.
But we need to separate the two ideas:
The risk lies in conflating the two.
CVs and LinkedIn profiles are riddled with bias, inconsistency, and omission. They’re self-authored, unverified, and often written strategically – for example, to enhance certain experiences or downplay others in response to a job ad.
And different groups represent themselves in different ways. Ahuja (2024) showed, for example, that male MBA graduates in India tend to self-promote more than their female peers. Something as simple as a longer LinkedIn ‘About’ section becomes a proxy for perceived competence.
Job titles are vague. Skill descriptions vary. Proficiency is rarely signposted. Even where systems draw on internal performance data, the quality is often questionable. Ratings tend to cluster (remember the year everyone got a ‘3’ at your org?) and can often reflect manager bias or company culture more than actual output.
The most advanced skill inference platforms use layered data: open web sources like job ads and bios, public databases like O*NET and ESCO, internal frameworks, even anonymised behavioural signals from platform users. This breadth gives a more complete picture, and the models powering it are undeniably sophisticated.
But sophistication doesn’t equal accuracy.
These systems rely heavily on proxies and correlations, rather than observed behaviour. They estimate presence, not proficiency. And when used in high-stakes decisions, that distinction matters.
In many inference systems, it’s hard to trace where a skill came from. Was it picked up from a keyword? Assumed from a job title? Correlated with others in similar roles? The logic is rarely visible, and that’s a problem, especially when decisions based on these inferences affect access to jobs, development, or promotion.
Inferred skills suggest someone might have a capability. But hiring isn’t about possibility. It’s about evidence of capability. Saying you’ve led a team isn’t the same as doing it well. Collecting or observing actual examples of behaviour allows you to evaluate someone’s true competence at a claimed skill.
Some platforms try to infer proficiency, too, but this is still inference, not measurement. No matter how smart the model, it’s still drawing conclusions from indirect data.
By contrast, validated assessments like structured interviews, simulations, and psychometric tools are designed to measure. They observe behaviour against defined criteria, use consistent scoring frameworks (like Behaviourally Anchored Rating Scales, or BARS), and provide a transparent, defensible basis for decision-making. In doing this, the level or proficiency of a skill can be placed on a properly calibrated scale.
But here’s the thing: we don’t have to choose one over the other.
The real opportunity lies in combining the rigour of measurement with the scalability of inference.
Start with measurement
Define the skills that matter. Use structured tools to capture behavioural evidence. Set a clear standard for what good looks like. For example, define Behaviourally Anchored Rating Scales (BARS) when assessing interviews for skills. Using a framework like Sapia.ai’s Competency Framework is critical for defining what you want to measure.
Layer in inference
Apply AI to scale scoring, add contextual nuance, and detect deeper patterns that human assessors might miss, especially when reviewing large volumes of data.
Anchor the whole system in transparency and validation
Ensure people understand how inferences are made by providing clear explanations. Continuously test for fairness. Keep human oversight in the loop, especially where the stakes are high. More information on ensuring AI systems are transparent can be found in this paper.
This hybrid model respects the strengths and limits of both approaches. It recognises that AI can’t replace human judgement, but it can enhance it. That inference can extend reach, but only measurement can give you higher confidence in the results.
Inference can support and guide, but only measurement can prove. And when people’s futures are on the line, proof should always win.
Ahuja, A. (2024). LinkedIn profile analysis reveals gender-based differences in self-presentation among Indian MBA graduates. Journal of Business and Psychology.
Hiring for care is unlike any other sector. Recruiters are looking for people who can bring empathy, resilience, and energy to the most demanding human roles. Whether it’s dental care, mental health, or aged care, new hires are charged with looking after others when they’re most vulnerable. The stakes are high.
Hiring for care is exactly where leveraging ethical AI can make the biggest impact.
The best carers don’t always have the best CVs.
That’s why our chat-based AI interview doesn’t screen for qualifications. It screens for the the skills that matter when caring for others. The traits that define a brilliant care worker, things like:
Empathy, Self-awareness, Accountability, Teamwork, and Energy.
The best way to uncover these traits is through structured behavioural science, delivered through an experience that allows candidates to open up. Giving candidates space to give real-life, open-text answers. With no time pressure or video stress. Then, our AI picks up the signals that matter, free from any demographic data or bias-inducing signals.
Candidates’ answers to our structured interview questions aren’t simply ticking boxes. They’re a window into how someone shows up under pressure. And they’re helping leading care organisations hire people who belong in care and those who stay.
Inclusivity should be a core foundation of any talent assessment, and it’s a fundamental requirement for hirers in the care industry.
When healthcare hirers use chat-based AI interviews, designed to be inclusive for all groups, candidates complete their interviews when and where they choose, without the bias traps of face-to-face or phone screening. There are no accents to judge, no assumptions, just their words and their story.
And it works:
Drop-offs are reduced, and engagement & employer brand advocacy go up. Building a brand that candidates want to work for includes providing a hiring experience that candidates want to complete.
Our smart chat already works for some of the most respected names in healthcare and community services. Here’s a sample of the outcomes that are possible by leveraging ethical AI, a validated scientific assessment, wrapped in an experience that candidates love:
The case study tells the full story of how Sapia.ai helped Anglicare, Abano Healthcare, and Berry Street transform their hiring processes by scaling up, reducing burnout, and hiring with heart.
Download it here:
A new study has just confirmed what many in HR have long suspected: traditional psychometric tests are no longer the gold standard for hiring.
Published in Frontiers in Psychology, the research compared AI-powered, chat-based interviews to traditional assessments, finding that structured, conversational AI interviews significantly reduce social desirability bias, deliver a better candidate experience, and offer a fairer path to talent discovery.
We’ve always believed hiring should be about understanding people and their potential, rather than reducing them to static scores. This latest research validates that approach, signalling to employers what modern, fair and inclusive hiring should look like.
While used for many decades in the absence of a more candidate-first approach, psychometric testing has some fatal flaws.
For starters, these tests rely heavily on self-reporting. Candidates are expected to assess their own traits. Could you truly and honestly rate how conscientious you are, how well you manage stress, or how likely you are to follow rules? Human beings are nuanced, and in high-stakes situations like job applications, most people are answering to impress, which can lead to less-than-honest self-evaluations.
This is known as social desirability bias: a tendency to respond in ways that are perceived as more favourable or acceptable, even if they don’t reflect reality. In other words, traditional assessments often capture a version of the candidate that’s curated for the test, not the person who will show up to work.
Worse still, these assessments can feel cold, transactional, even intimidating. They do little to surface communication skills, adaptability, or real-world problem solving, the things that make someone great at a job. And for many candidates, especially those from underrepresented backgrounds, the format itself can feel exclusionary.
Enter conversational AI.
Organisations have been using chat-based interviews to assess talent since before 2018, and they offer a distinctly different approach.
Rather than asking candidates to rate themselves on abstract traits, they invite them into a structured, open-ended conversation. This creates space for candidates to share stories, explain their thinking, and demonstrate how they communicate and solve problems.
The format reduces stress and pressure because it feels more like messaging than testing. Candidates can be more authentic, and their responses have been proven to reveal personality traits, values, and competencies in a context that mirrors honest workplace communication.
Importantly, every candidate receives the same questions, evaluated against the same objective, explainable framework. These interviews are structured by design, evaluated by AI models like Sapia.ai’s InterviewBERT, and built on deep language analysis. That means better data, richer insights, and a process that works at scale without compromising fairness.
The new study, published in Frontiers in Psychology, put AI-powered, chat-based interviews head-to-head with traditional psychometric assessments, and the results were striking.
One of the most significant takeaways was that candidates are less likely to “fake good” in chat interviews. The study found that AI-led conversations reduce social desirability bias, giving a more honest, unfiltered view of how people think and express themselves. That’s because, unlike multiple-choice questionnaires, chat-based assessments don’t offer obvious “right” answers – it’s on the candidate to express themselves authentically and not guess teh answer they think they would be rewarded for.
The research also confirmed what our candidate feedback has shown for years: people actually enjoy this kind of assessment. Participants rated the chat interviews as more engaging, less stressful, and more respectful of their individuality. In a hiring landscape where candidate experience is make-or-break, this matters.
And while traditional psychometric tests still show higher predictive validity in isolated lab conditions, the researchers were clear: real-world hiring decisions can’t be reduced to prediction alone. Fairness, transparency, and experience matter just as much, often more, when building trust and attracting top talent.
Sapia.ai was spotlighted in the study as a leader in this space, with our InterviewBERT model recognised for its ability to interpret candidate responses in a way that’s explainable, responsible, and grounded in science.
Today, hiring has to be about earning trust and empowering candidates to show up as their full selves, and having a voice in the process.
Traditional assessments often strip candidates of agency. They’re asked to conform, perform, and second-guess what the “right” answer might be. Chat-based interviews flip that dynamic. By inviting candidates into an open conversation, they offer something rare in hiring: autonomy. Candidates can tell their story, explain their thinking, and share how they approach real-world challenges, all in their own words.
This signals respect from the employer. It says: We trust you to show us who you are.
Hiring should be a two-way street – a long-held belief we’ve had, now backed by peer-reviewed science. The new research confirms that AI-led interviews can reduce bias, enhance fairness, and give candidates control over how they’re seen and evaluated.