MELBOURNE, July 2020: Australian AI recruitment start-up Sapia, has published peer-reviewed research validating a new AI-based approach to talent assessment that determines personality and job suitability through text.
The research was published by IEEE. https://ieeexplore.ieee.org/document/9121971
Personality assessments have long been used to supplement CV data. It is widely accepted that one’s personality can be a predictor of job performance and suitability. Thus, Sapia uses structured text-based interviews, NLP, and machine learning to identify personality traits by analysing text answers to questions related to the job being applied for.
Every candidate gets a “chat based smart interview”. As no demographic data is gathered from other sources such as CVs, the process is blind to gender, race and characteristics that are not relevant in candidate selection. The research validates the accuracy of Sapia’s AI approach. Lastly, it also signals a huge improvement to personality tests, where the candidate experience is underwhelming.
Also Know, Personality AI refers to the use of artificial intelligence (AI) technologies to analyze and understand human personality traits, tendencies, and behavior patterns. This field of AI has gained significant attention in recent years, as businesses and organizations seek to better understand their customers, employees, and other stakeholders.
Barbara Hyman, Sapia (Formerly PredictiveHire) CEO says chat-based interviews address the three big failures of current assessments – ghosting, bias and trust.
“Recruiters are the ultimate ghosters,” Ms Hyman says. “With Sapia, the fact that every single candidate receives a personalised learning profile is gold for candidates and your employer brand. Using text to analyse fit that’s blind to gender, race, age and any personal factors is a must-have in today’s current climate and means every company can introduce bias interruption for every hire and promotion. Imagine what that will do to diversity in hiring”
Principal Data Scientist Buddhi Jayatilleke says “language has long been seen as a source of truth for personality- it defines who we are. This technology offers a direct way to understand personality from language. All is done by using an experience that is human and empowering. Additionally, this capability can be used for assessment and personalised career coaching. Furthermore, it could be a game changer for job seekers, universities, and employers.”
Candidates across 34 countries have experienced Sapia’s unique chat-based interviews. More insight into how the technology works can be found here. https://sapia.ai/science-explained/
Sapia (Formerly PredictiveHire) is a team of data scientists, engineers and HR professionals. Together we have built a product suite that is based on science and built to humanise hiring. Sapia believes that relying on data to drive your most important decisions. Who you hire/ promote, enhances trust and confidence that decisions are fair. We also serve customers in the UK, South Africa, India Australia, and New Zealand.
To keep up to date on all things “Hiring with Ai” subscribe to our blog!
Finally, you can try out Sapia’s Chat Interview right now, or leave us your details to get a personalised demo.
You and I both know that adding more headcount will not help the issue [of recruiters being overworked], since it’ll just result in more people doing more tasks.
At one point, we had General Motors in a position where we were having quarterly go-to-market meetings every quarter.
As a leadership team, we met to determine what we wanted to achieve in the next quarter and what it would take to get there.
When I started running the go-to-market functions for my boss, Cyril George, I told him that no one here knew what their KPIs were because it wasn’t clear; it was like everything was on fire all the time.
So we began having these go-to-market meetings, and a significant portion of them focused on the tech and innovation that we were driving to resource the team.
Then someone asked, “What’s the point once we implement all of this?”
I knew the subtext was, “Are we laying people off? Are we getting rid of recruiters?”
I responded, “No, the point is for you not to be working 65 hours a week every week.”
The room fell silent; there was no slow clap, just disbelief and shock.
They thought, “I don’t think that’s real,” but it is.
That’s what tech can do, you know.
Not only can it help for one quarter, but it can also make a difference for years to come.
So, stop thinking of buying tech for new best practices that it can bring, and start thinking of it as a way to extend our capacity sustainably and meaningfully.
Yeah, I see it the same way, in terms of giving you leverage.
Every time you hire someone for your team, you gain more leverage, allowing you to achieve more.
Technology does the same thing, but on a larger scale.
Listen to the full episode of our podcast featuring Kyle Lagunas here:
We hope that the debate over the value of diverse teams is now over. There is plenty of evidence that diverse teams lead to better decisions and therefore, business outcomes for any organisation.
This means that CHROs today are being charged with interrupting the bias in their people decisions and expected to manage bias as closely as the CFO manages the financials.
But the use of Ai tools in hiring and promotion requires careful consideration to ensure the technology does not inadvertently introduce bias or amplify any existing biases.
To assist HR decision-makers to navigate these decisions confidently, we invite you to consider these 8 critical questions when selecting your Ai technology.
You will find not only the key questions to ask when testing the tools but why these are critical questions to ask and how to differentiate between the answers you are given.
Another way to ask this is: what data do you use to assess someone’s fit for a role?
First up- why is this an important question to ask …
Machine-learning algorithms use statistics to find and apply patterns in data. Data can be anything that can be measured or recorded, e.g. numbers, words, images, clicks etc. If it can be digitally stored, it can be fed into a machine-
The process is quite basic: find the pattern, apply the pattern.
This is why the data you use to build a predictive model, called training data, is so critical to understand.
In HR, the kinds of data that could be used to build predictive models for hiring and promotion are:
If you consider the range of data that can be used in training data, not all data sources are equal, and on its surface, you can certainly see how some carry the risk of amplifying existing bases and the risk of alienating your candidates.
Using data that is invisible to the candidate may impact your employer brand. And relying on behavioural data such as how quickly a candidate completes the assessment, social data or any data that is invisible to the candidate might expose you to not only brand risk but also a legal risk. Will your candidates trust an assessment that uses data that is invisible to them, scraped about them or which can’t be readily explained?
Increasingly companies are measuring the business cost from poor hiring processes that contribute to customer churn. 65% of candidates with a positive experience would be a customer again even if they were not hired and 81% will share their positive experience with family, friends and peers (Source: Talent Board).
Visibility of the data used to generate recommendations is also linked to explainability which is a common attribute now demanded by both governments and organisations in the responsible use of Ai.
Video Ai tools have been legally challenged on the basis that they fail to comply with baseline standards for AI decision-making, such as the OECD AI Principles and the Universal Guidelines for AI.
Or that they perpetuate societal biases and could end up penalising nonnative speakers, visibly nervous interviewees or anyone else who doesn’t fit the model for look and speech.
If you are keen to attract and retain applicants through your recruitment pipeline, you may also care about how explainable and trustworthy your assessment is. When the candidate can see the data that is used about them and knows that only the data they consent to give is being used, they may be more likely to apply and complete the process. Think about how your own trust in a recruitment process could be affected by different assessment types.
1st party data is data such as the interview responses written by a candidate to answer an interview question. It is given openly, consensually and knowingly. There is full awareness about what this data is going to be used for and it’s typically data that is gathered for that reason only.
3rd party data is data that is drawn from or acquired through public sources about a candidate such as their Twitter profile. It could be your social media profile. It is data that is not created for the specific use case of interviewing for a job, but which is scraped and extracted and applied for a different purpose. It is self-evident that an Ai tool that combines visible data and 1st party data is likely to be both more accurate in the application for recruitment and have outcomes more likely to be trusted by the candidate and the recruiter.
At PredictiveHire, we are committed to building ethical and engaging assessments. This is why we have taken the path of a text chat with no time pressure. We allow candidates to take their own time, reflect and submit answers in text format.
We strictly do not use any information other than the candidate responses to the interview questions (i.e. fairness through unawareness – algorithm knows nothing about sensitive attributes).
For example, no explicit use of race, age, name, location etc, candidate behavioural data such as how long they take to complete, how fast they type, how many corrections they make, information scraped from the internet etc. While these signals may carry information, we do not use any such data.
Another way to ask this is – Can you explain how your algorithm works? and does your solution use deep learning models?
This is an interesting question especially given that we humans typically obfuscate our reasons for rejecting a candidate behind the catch-all explanation of “Susie was not a cultural fit”.
For some reason, we humans have a higher-order need and expectation to unpack how an algorithm arrived at a recommendation. Perhaps because there is not much to say to a phone call that tells you were rejected for cultural fit.
This is probably the most important aspect to consider, especially if you are the change leader in this area. It is fair to expect that if an algorithm affects someone’s life, you need to see how that algorithm works.
Transparency and explainability are fundamental ingredients of trust, and there is plenty of research to show that high trust relationships create the most productive relationships and cultures.
This is also one substantial benefit of using AI at the top of the funnel to screen candidates. Subject to what kind of Ai you use, it enables you to explain why a candidate was screened in or out.
This means recruitment decisions become consistent and fairer with AI screening tools.
But if Ai solutions are not clear why some inputs (called “features” in machine learning jargon) are used and how they contribute to the outcome, explainability becomes impossible.
For example, when deep learning models are used, you are sacrificing explainability for accuracy. Because no one can explain how a particular data feature contributed to the recommendation. This can further erode candidate trust and impact your brand.
The most important thing is that you know what data is being used and then ultimately, it’s your choice as to whether you feel comfortable to explain the algorithm’s recommendations to both your people and the candidate.
Assessment should be underpinned by validated scientific methods and like all science, the proof is in the research that underpins that methodology.
This raises another question for anyone looking to rely on AI tools for human decision making – where is the published and peer-reviewed research that ensures you can have confidence that a) it works and b) it’s fair.
This is an important question given the novelty of AI methods and the pace at which they advance.
At PredictiveHire, we have published our research to ensure that anyone can investigate for themselves the science that underpins our AI solution.
We continuously analyse the data used to train models for latent patterns that reveal insights for our customers as well as inform us of improving the outcomes.
It’s probably self-evident why this is an important question to ask. You can’t have much confidence in the algorithm being fair for your candidates if no one is testing that regularly.
Many assessments report on studies they have conducted on testing for bias. While this is useful, it does not guarantee that the assessment may not demonstrate biases in new candidate cohorts it’s applied on.
The notion of “data drift” discussed in machine learning highlights how changing patterns in data can cause models to behave differently than expected, especially when the new data is significantly different from the training data.
Therefore on-going monitoring of models is critical in identifying and mitigating risks of bias.
Potential biases in data can be tested for and measured.
These include all assumed biases such as between gender and race groups that can be added to a suite of tests. These tests can be extended to include other groups of interest where those group attributes are available like English As Second Language (EASL) users.
On bias testing, look out for at least these 3 tests and ask to see the tech manual and an example bias testing report.
At PredictiveHire, we conduct all the above tests. We conduct statistical tests to check for significant differences between groups of feature values, model outcomes and recommendations. Tests such as t-tests, effect sizes, ANOVA, 4/5th, Chi-Squared etc. are used for this. We consider this standard practice.
We go beyond the above standard proportional and distribution tests on fairness and adhere to stricter fairness considerations, especially at the model training stage on the error rates. These include following guidelines set by IBM’s AI Fairness 360 Open Source Toolkit. Reference: https://aif360.mybluemix.net/) and the Aequitas project at the Centre for Data Science and Public Policy at the University of Chicago
We continuously analyse the data used to train models for latent patterns that reveal insights for our customers as well as inform us of improving the outcomes.
We all know that despite best intentions, we cannot be trained out of our biases. Especially the unconscious biases.
This is another reason why using data-driven methods to screen candidates is fairer than using humans.
Biases can occur in many different forms. Algorithms and Ai learn according to the profile of the data we feed it. If the data it learns from is taken from a CV, it’s only going to amplify our existing biases. Only clean data, like the answers to specific job-related questions, can give us a true bias-free outcome.
If any biases are discovered, the vendor should be able to investigate and highlight the cause of the bias (e.g. a feature or definition of fitness) and take corrective measure to mitigate it.
If you care about inclusivity, then you want every candidate to have an equal and fair opportunity at participating in the recruitment process.
This means taking account of minority groups such as those with autism, dyslexia and English as a second language (EASL), as well as the obvious need to ensure the approach is inclusive for different ethnic groups, ages and genders.
At PredictiveHire, we test the algorithms for bias on gender and race. Tests can be conducted for almost any group in which the customer is interested. For example, we run tests on “English As a Second Language” (EASL) vs. native speakers.
If one motivation for you introducing Ai tools to your recruitment process is to deliver more diverse hiring outcomes, it’s natural you should expect the provider to have demonstrated this kind of impact in its customers.
If you don’t measure it, you probably won’t improve it. At PredictiveHire, we provide you with tools to measure equality. Multiple dimensions are measured through the pipeline from those who applied, were recommended and then who was ultimately hired.
8. What is the composition of the team building this technology?
Thankfully, HR decision-makers are much more aware of how human bias can creep into technology design. Think of how the dominance of one trait in the human designers and builders have created an inadvertent unfair outcome.
In 2012, YouTube noticed something odd.
About 10% of the videos being uploaded were upside down.
When designers investigated the problem, they found something unexpected: Left-handed people picked up their phones differently, rotating them 180 degrees, which lead to upside-down videos being uploaded,
The issue here was a lack of diversity in the design process. The engineers and designers who created the YouTube app were all right-handed, and none had considered that some people might pick up their phones differently.
In our team at PredictiveHire, from the top down, we look for diversity in its broadest definition.
Gender, race, age, education, immigrant vs native-born, personality traits, work experience. It all adds up to ensure that we minimise our collective blind spots and create a candidate and user experience that works for the greatest number of people and minimises bias.
What other questions have you used to validate the fairness and integrity of the Ai tools you have selected to augment your hiring and promotion processes?
We’d love to know!
You can try out PredictiveHire’s FirstInterview right now, or leave us your details to get a personalised demo
To find out how to improve candidate experience using Recruitment Automation, we also have a great eBook on candidate experience.
By Jennifer Hewett, Australian Financial Review, 31 January
The online questionnaire wants to know whether I respect and comply with authority. I get five options – strongly agree, agree, neutral, disagree or strongly disagree. I tick “neutral”. Well sort of, sometimes, I think to myself.
Same choice for whether I am good at finding fault with what’s around me at work. I tick “neutral” again, guiltily acknowledging it’s just possible my editor might have a different opinion about whether I am far too good at that particular skill.
The choice seems less ambiguous when I am asked whether I forget to put things back in their proper place. I hover over “strongly agree” or “agree” and tick the latter – perhaps a little optimistically.
And on it goes for 90 questions, with slight variations in the possible answers, as devised by an AI (artificial intelligence) algorithm. My responses to the bot will determine whether I get to the next stage of actually being interviewed for a job by a real person. AI approves who you should interview
I soon get an encouraging email from Michael Morris, chief executive of Employsure – a company which provides advice on workplace relations and health and safety issues to small businesses. If I ever give up journalism, Morris tells me, I can try for a new career at Employsure. AI has approved me. Despite my deep scepticism about the process, I can’t help but feel a little pleased by the bot’s assessment.
That is because my rather self-serving answers to random personality questions fit those of the best performers at Employsure. There’s no possibility of ageism or sexism or any other latest “ism” influencing that. No old schoolmates or university or sporting framework, no biases about looks or clothes or mannerisms or personal history.
Instead, I participated in what is a variation on a personality test – based on the algorithmic analysis provided by another company, Sapia, operating in Europe and Australia and with 20 clients.
Morris says Employsure tested the performance of employees selected by Sapia’s algorithm against the choices of Employsure’s own human recruitment team for much of last year.
The fast-growing company hired around 450 people in 2018 with a workforce now totalling more than 800. Morris wanted good people and those more likely to stay.
The experience convinced him that rather than using more traditional CVs to screen applicants, it was worth paying Sapia for its AI technology as Employsure continues to expand its numbers this year. Employsure now only interviews the 10-15 per cent of those who are graded “yes” or “maybe” by the bot.
“The overlay of AI made a significant difference in overall performance, productivity and tenure,” Morris says. “And it means the recruitment team can have a head start on engaging in better conversations with those who have interviews.” This is still a distinct minority view among Australian businesses which have been generally reluctant to embrace the promise of AI when it comes to hiring.
Read: The Ultimate Guide to Interview Automation
The trend to make greater use of AI in business generally is inevitable and accelerating. Just consider all those online “conversations” we now have about customer service and products as the ever-patient bot nudges us this way and that.
Just as inevitably, it is leading to community concerns about whether AI will be used to replace too many people’s jobs. According to a study by the McKinsey Global Institute, intelligent agents and robots could eliminate as much as 30 per cent of human labour by 2030. The scale would dwarf the move away from agricultural labour during the 1900s in the United States and Europe.
Of course, the record of technology shifts over centuries always ending up creating many more other types of jobs does not completely soothe fears that this time it’s different. Even if such alarm is overstated, dramatic changes in technology can certainly prove socially and economically disruptive for long periods. AI can also be scary.
But this version of AI is more about filling new jobs more efficiently. Many large global companies already use it to filter job applicants, especially those coming in at lower levels. Its advocates argue it efficiently eliminates bias or the tendency for people to hire in their own image.
Not that this always goes smoothly – even for the most digitally sophisticated businesses. Amazon abandoned its own AI hiring tool last October when management realised it had only introduced more bias into the process. Its AI system was based on modelling the CVs of those already at the company – who tended to be male. Naturally, that made prospective hires more likely to be male too. So much for gender-diversity targets.
Sapia’s chief executive is Barb Hyman, formerly a human resources executive for the online real estate advertising company REA Group. She says the system doesn’t work for those companies that don’t measure the performance of their existing employees but the data becomes more and more accurate as more information is added.
By matching responses of applicants against only those employees who are already doing well, it can be extremely efficient with immediate payback – especially for larger companies. The data can also be used to change the culture in an organisation by screening the types of personalities who are hired.
Not surprisingly, Hyman says the data demonstrates how different personalities are better fitted to different sorts of roles. So those who do well in caring jobs tend to be reliable and demonstrate traits of modesty and humility. Good salespeople are focused, somewhat self-absorbed, disorganised and transactional. Those who are involved in building long-term business relationships need to be more adaptable, resilient and open.
Sounds more like common sense than AI. But there’s less and less of that around anywhere. AI beckons instead.