Tomas Chamorro-Premuzic is a renowned organizational psychologist who frequently speaks and writes on topics at the intersection of talent, innovation and AI. Today, he is a Professor of Business Psychology at Columbia University and University College of London, and the Chief Talent Scientist and at ManpowerGroup, a staffing firm. Follow him on Twitter @drtcp.
It’s true that well-designed, structured interview processes are predictive of future behavior, including job performance—but they rarely happen. Most interviews are like a first date; the behavior in the first meeting doesn’t necessarily reflect what you see months or years later.
Between two organizations, the one that is more data driven will be better able to spot trends and talent, and they can be more meritocratic. Because of this, we can expect companies with data-driven hiring practices to maximize their talent ROI and outperform competitors.
There’s now more digital data available on candidates and employees than ever before, from social profiles, email, video interviews and more. With AI, companies can mine this data for predictive and meaningful insights on a person’s potential in the workplace. This opens up unprecedented opportunities to help organizations deploy people in the best possible ways.
Some people are put off by the idea of automation in talent decision-making and can feel creeped out by the thought of algorithms evaluating their resume or social media presence. To mitigate reluctance, Tomas believes companies should lead with transparency; it should be clear exactly what personal information the company captures and how automation will be used, and candidates should be allowed to opt in or out.
Not only will this benefit organizations by encouraging acceptance but it will also help to educate candidates on what their personal data actually says about them and where their skills could best fit within an organization.
Human bias has pervaded hiring throughout history. Moving forward, Tomas believes that we must decrease reliance on human observers and instead use technology, including AI, to identify the predictive signals of somebody’s talent or potential without weighing attributes like ethnicity, gender and age. Ultimately, the use of AI in talent practices will lead to more merit based decision making.
EPISODE 11: TOMAS CHAMORRO-PREMUZIC
Unlocking Employee Potential with AI
Jim Freeze Hi and welcome. This is Jim Freeze, your host, and this is The ConversAItion, a podcast airing viewpoints on the impact of artificial intelligence on business and society.
The ConversAItion is presented by Interactions, a conversational AI company, that builds intelligent virtual assistants, capable of human level communication and understanding. In today’s episode, we’re talking about how AI can improve hiring practices.
Joining us to discuss the shortcomings of traditional hiring and how AI can help is Dr. Tomas Chamorro-Premuzic. Tomas is an organizational psychologist focused on personality profiling, people analytics and leadership development. He’s a prolific speaker and author. As a matter of fact, he has written 10 books and more than 150 scientific papers and often contributes to publications like Harvard Business Review, Fast Company and Forbes.
Today, Tomas is Professor of Business Psychology at Columbia University and Chief Talent Scientist at staffing firm, Manpower Group. Tomas, welcome to The ConversAItion.
Tomas Chamorro-Premuzic Hi, thank you for having me.
Jim Freeze We’re thrilled that you’re here. So, throughout your career, you’ve explored a wide range of topics related to personality, talent, and leadership. When did you become interested in the intersection of artificial intelligence and talent?
Tomas Chamorro-Premuzic So I think probably around 10 years ago, and up until that time, I had done a lot of research and commercial or real world work, exploring the accuracy of traditional psychological assessment tools to identify future leaders. Determine whether somebody has talent for an area or not. The more conventional area of application of personality assessments and talent identification tools in the world of HR.
Then, of course with the explosion of smartphones and people producing so much data and the what was then called big data revolution. That made me think that maybe AI could help us scale the science of personality and assessment to a whole new dimension.
I think that the science of psychological assessment is really valuable to help organizations understand their people and people understand themselves. Most of the problems that organizations have today have to do with a crisis of understanding; not understanding where to deploy their talent, what people are good at, what they’re not interested in. And simultaneously you have people who are not aware of what their potential and their talents are. So that’s why I think AI can really amplify and help us scale what I think it’s a very powerful and well established science.
Jim Freeze Yeah. And as I hear you articulate that and I’m thinking about it on the front end of the process when you’re actually interviewing people and you’ve discussed the shortcomings of traditional hiring practices, like unstructured job interviews. Can you elaborate a little bit on the problems with the way businesses have historically recruited and hired?
Tomas Chamorro-Premuzic I think the main problem is that even though you can design very rigorous and predictive interviews, it suffers from a dilution problem. Most people feel that they are natural judges of other people’s characters and that they don’t have to work very hard to design a very standardized, methodical and structured process. If you think about it, most of the interview experiences people have are still very informal, very unstructured, and there’s not much homework or much data underpinning this process. And even though we’ve known for many, many years that well-designed rigorous interviews, structured job interviews are very predictive of future behavior, including job performance, they rarely happen.
And for sure people overestimate the importance that interview behaviors have as indicators of somebody’s potential. What you see in an interview is the equivalent of what you would see on a first date when you go on a romantic date. You see a person’s best behavior, and that will not extrapolate to what you see six months later, or six years later, if you marry that person.
Same happens with the employees. And all the things that we want to pay attention to during an interview, whether it’s face to face or virtual, are the things that fair societies are trying desperately to ignore right now. Information about how attractive you are, what ethnicity you have, what gender or age you have, or your group you are from. So, the only way to address that is to actually remove, or de-emphasize reliance on human observers and use technology, including AI, to identify or pick up the predictive signals of somebody’s talent or potential, while actually ignoring all the things that are uncomfortable and often unethical demographic categories or groups.
Jim Freeze Can you talk about some of the emerging technologies or applications of AI that are starting to—at least as you’ve observed it—to transform recruiting and hiring practices?
Tomas Chamorro-Premuzic Yes. I think that I would use AI mostly in this phase of HR or talent would be kind of equate to machine learning or deep learning algorithms. And it’s a method for treating data and translating data into insights. Insights that are predictive and that can tell you whether somebody is a good match for a certain job role or career. Where there is a little bit more variety and where you can talk about the innovations is in the types of data capture or the types of signals that we are now able to mine with AI.
So, it would include things like your social network or your social media footprint. The internal exhaust of data that you leave within your job organization, or employer, email, metadata, content, context.
If you’re looking at video or digital interview technology, there’s so many signals. So, of course there’s natural language processing, which I know you’ve covered in your show before. So translating some of the words people say into aspects of their character personality, or potential. Their body language, physical properties of speech. There’s now more digital data available than ever before. And of course, more predictive and meaningful digital signals on somebody’s potential, than what you can see in the physical or analog world. And that opens up unprecedented opportunities to understand what somebody’s potential is and to help organizations deploy people in the best possible ways.
Jim Freeze Yeah. So, that kind of data and the application of AI could really help companies make better, more accurate hiring decisions.
Tomas Chamorro-Premuzic Correct. And I think it’s important to acknowledge that there is a lot of room for improvement. The baseline right now is not very high. Most people choose careers or jobs in a very serendipitous way. Then it’s too late to change. Most people are not engaged at work. Most people don’t understand what their best potential would be or how they can develop their career potential. And employers have a very rudimentary level of understanding of their people. It takes a manager with great people skills and high EQ and a lot of social skills and great motivation to know their employees and their team members.
And by the way, that you can’t even scale, if that person has a team of more than 20 people or a lot of reports it’s just not feasible for them to have a good understanding of what people are like and what makes them tick. So, that’s where technology comes in. Technology is always about doing more with less. And in this case, it’s identifying the critical markers or signals that make you who you are and different from others when it comes to work related skills.
Jim Freeze We have more data than ever at our disposal. And you talked about some of the things like social profiles where you can grab information on candidates. Companies can create a fairly comprehensive candidate profile, but I think you’ve argued that there’s a difference between what we can know and what we should know about candidates. What types of data should be leveraged in making hiring decisions and how can companies ensure that they don’t cross the line into an invasive use of data?
Tomas Chamorro-Premuzic Yeah, exactly. And I’m really interested in the ethical side of this, because the legal side at the end of the day, our listeners might be in different places and each country, and sometimes each state has different legalities or processes to regulate this. But the ethical side of things actually has not changed that much. And we can safely assume that it can be applied universally to different places. So, I believe that yes, there is today a difference between what we could know and what we should know about people.
But I believe that if you enable people to opt in, you obtain informed consent. You explain to them what data are being captured and what’s being done to that data or with that data. And then finally, you invite them to either share that information or have others, recruiters or employers use that information. I believe that’s a very transparent and ethical way of leveraging these innovations.
Jim Freeze Yeah, as you’re talking about that, I’m thinking about how there are people who’ve made a business of helping people build profiles online, like on LinkedIn to make them more attractive candidates. I was just thinking about that’s a really interesting notion that, you’re pulling this data, but you’re also then potentially providing it to potential candidates say, look, this is what your social profile says. If you’re happy with that, then great. If you’re not, then maybe you need to think about making some changes. So, in that sense, it’s almost advisory to candidates as well.
Tomas Chamorro-Premuzic Exactly. And I think it’s important to understand that all that is very positive, right? So, before the whole area of digital talent identification or talent management really came, which is relatively recent, 10, 15 years ago. If you look for example, in our business within Manpower Group, our recruiters had worked with candidates for decades. Helping them understand how to dress for an interview, how to present, what employers want to hear. And that’s not a bad thing. That’s not an incentive or shouldn’t be seen as, Oh, we’re trying to game a system or help them cheat. We’re just trying to display that right etiquette and fit in or display or good organizational citizenship, which is part of talent.
Jim Freeze Interesting. So, I’m thinking back to our first season of the podcast. We had a gentleman on, by the name of Gabriel Skantze, who was the head of a company called Furhat Robotics. They make these human-like robots and he believed that a great application for their technology was conducting unbiased first round interviews. He thought that was a great early application for AI,that they could be very objective, very similar to what you’ve said.
But interestingly enough, I think you’ve written that humans often trust their gut over data-driven recommendations. What do you think it takes for organizations to truly make the shift to a data driven decision making process, as opposed to a gut feel as it relates to hiring?
Tomas Chamorro-Premuzic Yeah. So, mostly it requires some humility, some curiosity, some self criticism. And of course it requires some understanding of the data and technology. It is clear to me that it is not always possible to persuade people of something, even when you show them facts and data. But at the end of the day, what we need to understand is that organizations that have the humility and the curiosity to make decisions, following the data, even when they run counter their own instincts, can be expected to outperform their competitors. Between two organizations, the one that is more data driven, will be better able to spot trends, better able to spot and attract talent. Therefore we can expect them to outperform their competitors because that’s the ROI to having more talented employees. And in particular, more talented leaders.
Imagine a company that only selects leaders because they get along with the people in an interview, or they hire people in a nepotistic way. You can expect that company to be less meritocratic and you can expect that company to perform less than a more meritocratic alternative. So, at the end of the day, if data and AI can help companies be more meritocratic and more talent-centric, there will be an ROI.
Jim Freeze Have you seen over the last year or two less pushback from candidates and or companies that are doing the recruiting as, is there a growing acceptance of the notion that AI has a really critical role to play in talent recruitment?
Tomas Chamorro-Premuzic Yes. And in particular from the organizational side: the hiring managers, employers, HR departments. Most big companies have within their HR function now a people analytics line, with data scientists, and they’re trying to really follow the data and be more fact and evidence-based. That shows you, and this is a fast, fast growing field that shows you that openness to AI and data in the field of HR and talent is advancing rapidly.
Where I don’t see equivalent progress is on the candidate side. And I think even some organizations don’t implement certain things like video interviewing or digital interviewing or algorithmic selection or deselection because they fear a backlash from the PR standpoint. And I understand that they’re cautious and they don’t want to be seen as doing something that is creepy or big brother-like, et cetera. But they also have the responsibility to educate the market and especially candidates.
I think candidates have double standards and they’re being very nice and forgiving with all the human biases that have contaminated the hiring practices for so long. Candidates love the job interview, even though it lends itself to unconscious and conscious biases and even though so many interviewers are ageist, sexist, racist, and have all these prejudices that they cannot suddenly switch off. And yet there are all these criticisms because occasionally companies have tried to implement AI based or algorithmic hiring that has produced biased outcomes. But only because the algorithms were trained to imitate or emulate human preferences, right?
So that’s what happened, for example, when chatbots or video interview algorithms were trained to predict who gets promoted in a certain company. Yes, it tended to be middle aged white males, but that doesn’t mean the algorithms were biased. It means there were successfully exhibiting or uncovering a bias that was there in the first place and doesn’t go away just because you don’t use AI. So, I think that’s quite interesting because AI, in a way is just a magnifying glass or lens or an X-ray that enables you to go into organizations to reveal biases. And we’re pointing the finger at it saying, “Ooh, AI is biased.” As if algorithms could be sexist or racist. As if AI and algorithms and computers had a fragile self esteem they need to protect by bringing other people down.
Jim Freeze Well, that’s a great way to close out this episode of The ConversAItion because I think you’re hitting on one of the things that we try to do here, which is to air viewpoints on the impact of artificial intelligence on business and society. Both the positive and the negative and I think what you just articulated was a misunderstanding that people have about artificial intelligence and how its proper use can actually help improve society. So, I think that was fantastic. And Tomas, thank you very much for being on The ConversAItion.
Tomas Chamorro-Premuzic Thank you so much for having me.
Jim Freeze On the next episode of The ConversAItion, we’ll explore how AI is disrupting the parking industry with Elan Mosbacher, the Senior Vice President of Strategy and Operations at SpotHero, the largest digital parking company in North America.
This episode of The ConversAItion was recorded remotely, and produced by Interactions, a Boston-based conversational AI company.
Well, that’s it for today’s ConversAItion. This is Jim Freeze, signing off, and we’ll see you next time.