Large Language Models, or LLMs, are the engines that have driven some of the biggest technological advancements in recent years. OpenAI’s GPT-4, Google’s Bard, and Metas Llama are examples of LLMs that power interfaces like ChatGPT, which has changed how many of us work and manage aspects of our daily lives.
At Sapia.ai, we use responsible AI to help organizations find the people who belong with their brands. Chat Interview™ enables our customers to interview every candidate, widening their talent pools and creating a more efficient, effective, and fair recruitment process.
With the advent of LLMs with enhanced language understanding and reasoning, there is a promising opportunity to extend these models in the recruitment space. However, using LLMs for activities like interview scoring demands not only innovation but also a strong commitment to responsible use and ethics. Any use of technology to make hiring more efficient should also make it fairer and transparent.
This blog post summarises the paper that we presented at SIOP 2024, titled “Responsible Use of Large Language Models for Response Grading and Explanations in Structured Interviews”, written by Yimeng Dai, Leo Pham, Ashlie Plants, and Buddhi Jayatilleke.
Our goal was to create an LLM that could grasp the unique nuances of interview conversations, leading to the creation of InterviewLLM. This domain-specific model was trained on a rich dataset from our Chat Interview™ platform, encompassing over 1.3 billion words from diverse industries, job roles, and geographies. The result is a model that can understand and generate interview dialogues with fewer possible stereotypes.
Building on the capabilities of InterviewLLM, we developed Saige™ (Sapia AI for Interview Grading and Explanations), a tool that takes interview grading to new heights. Saige™ is fine-tuned with a grading rubric crafted by our subject matter experts, employing Behaviorally Anchored Rating Scales (BARS) to ensure that assessments of candidate responses are anchored in the description of observable behaviors. By adopting BARS, the gold standard for creating rubrics for scoring interviews by humans, Saige™ allows human SMEs to add new competencies to be rated by Saige™, which most of the current mainstream automated interview grading approaches lack.
We didn’t stop at creating a more intelligent grading system with extensibility and explainability. Using LLMs in interviews demands a strong ethical commitment. We recognized the importance of addressing potential biases in AI, which led us to implement a de-biasing process during Saige™’s training. By combining the expertise of our human SMEs with AI feedback, we created a model that not only delivers accurate grades but does so in a manner that is fair and objective.
Saige™ also leverages LLM to explain why a candidate received a certain score, referencing actual examples from the candidate’s response and a description of how it exemplifies or fails to exemplify, the characteristics of the attribute that were defined by the SMEs. The characteristics of each attribute are defined by not only the many facets of them but also the frequency and depth in which they are demonstrated.
The introduction of Saige™ marks a significant milestone in the recruitment industry. In testing, Saige™ has demonstrated exceptional performance compared to LLaMA-2, with a higher agreement and correlation with SME grading.
The implications of Saige™ extend beyond just improved interview grading accuracy. By reducing biases in the grading process, we’re contributing to a more equitable hiring process, where candidates are evaluated based on their merits rather than unconscious prejudices.
Saige™’s ability to provide clear, natural language explanations for grading decisions creates a level of transparency not seen before in automated interviews. With the explanations, candidates can gain insights into areas for improvement and identify the attributes or abilities necessary for success in their careers; while hiring teams can learn more about each candidate and have confidence in their hiring decisions.
We’re at the start of our journey with InterviewLLM and Saige™. The opportunity to use this capability beyond assessment, to provide effective coaching to candidates and employees at scale is immense and exciting. It could enable them to understand how they can improve, fostering continuous growth and professional development within the organization. As we continue to refine these Generative AI tools and explore new applications of LLMs in recruitment, we remain committed to our core principles of fairness and transparency.
Why neuroinclusion can’t be a retrofit and how Sapia.ai is building a better experience for every candidate.
In the past, if you were neurodivergent and applying for a job, you were often asked to disclose your diagnosis to get a basic accommodation – extra time on a test, maybe the option to skip a task. That disclosure often came with risk: of judgment, of stigma, or just being seen as different.
This wasn’t inclusion. It was bureaucracy. And it made neurodiverse candidates carry the burden of fitting in.
We’ve come a long way, but we’re not there yet.
Over the last two decades, hiring practices have slowly moved away from reactive accommodations toward proactive, human-centric design. Leading employers began experimenting with:
But even these advances have often been limited in scope, applied to special hiring programs or specific roles. Neurodiverse talent still encounters systems built for neurotypical profiles, with limited flexibility and a heavy dose of social performance pressure.
Hiring needs to look different.
Truly inclusive hiring doesn’t rely on diagnosis or disclosure. It doesn’t just give a select few special treatment. It’s about removing friction for everyone, especially those who’ve historically been excluded.
That’s why Sapia.ai was built with universal design principles from day one.
Here’s what that looks like in practice:
It’s not a workaround. It’s a rework.
We tend to assume that social or “casual” interview formats make people comfortable. But for many neurodiverse individuals, icebreakers, group exercises, and informal chats are the problem, not the solution.
When we asked 6,000 neurodiverse candidates about their experience using Sapia.ai’s chat-based interview, they told us:
“It felt very 1:1 and trustworthy… I had time to fully think about my answers.”
“It was less anxiety-inducing than video interviews.”
“I like that all applicants get initial interviews which ensures an unbiased and fair way to weigh-up candidates.”
Some AI systems claim to infer skills or fit from resumes or behavioural data. But if the training data is biased or the experience itself is exclusionary, you’re just replicating the same inequity with more speed and scale.
Inclusion means seeing people for who they are, not who they resemble in your data set.
At Sapia.ai, every interaction is transparent, explainable, and scientifically validated. We use structured, fair assessments that work for all brains, not just neurotypical ones.
Neurodiversity is rising in both awareness and representation. However, inclusion won’t scale unless the systems behind hiring change as well.
That’s why we built a platform that:
Sapia.ai is already powering inclusive, structured, and scalable hiring for global employers like BT Group, Costa Coffee and Concentrix. Want to see how your hiring process can be more inclusive for neurodivergent individuals? Let’s chat.
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: