Video Interview Platform | Searchie

Video interviews can reduce bias in recruitment

Welcome back to another Searchie session!

This week we looked at how video interviews can affect bias in the recruitment process.

It’s no surprise that personal biases leak into the recruitment process. We explore examples from the real world that indicate a preference for hiring in the big technology companies.

Virtual video interview tools like Searchie offer recruiters the opportunity to get to know candidates beyond the CV. 


Recruitment consultants and in house recruiters work from a job description to help them stay focused on the traits they are looking for. Most often, they use the CV to benchmark how well a candidate fits to the role. However, research shows that past work experience is not always a good indicator of future job fit.

All our presentations are available on slideshare, including this one.

We’ve also included this presentation below so you can check it out.

[slideshare id=231951132&doc=overcomingbiasindataacquisition-200414052529&w=500]

Virtual Video Interviews Can Help Reduce Bias In The Recruitment Process [Transcript]

Hello, welcome back everybody.

Thank you very much for sharing your time with us on the stream today. We’ve got a topic which should really be surprising to my mind. It’s really kind of understand. The decision making is often influenced by sum of our experiences and some of the ways that we are using virtual video interviews to help recruiters hire people.

So we’re going to be talking about bias in Recruitment and artificial intelligence and kind of some of the the the trends that we can see that suggests that decision making process in a hiring decisions on always as objective as they perhaps believed to be as as as kind of computing instruments were very bad at being objective, right?

We like to Launch control on our experiences would like to review the world through the counter color of the lens with which we look at it. And sometimes that means the decision-making is kind of college by our emotions or experiences that we have regularly thinking about is in an age where we trying to teach machines how to how to think what are the lessons that we wanted to understand which is why the buys question keeps coming to my mind because we spent so much time trying to explore like what are the things that we are teaching these machines and how can I impact the people the other using the tools that we develop or exposed to them so that we can try and make it as Fair?

Can I use that word very subjective them very Loosely because this is really again designed by how you look at the problem. When I look at the situation with some of the things that we’re seeing it and then rather than just keeping already kind of ethereal. I want to try making love with tangible and trying to show you guys some of the stuff that you might be able to identify.

If you stop looking for the data of surprised. We are producing more data than we’ve ever previously in order of human history and facts. So if you look for the day, so you’d be surprised at the kind of stuff that you can learn from it, but there’s also a danger or not when we’re looking at some of the Condor of Decision making of the future is very dangerous to take an aggregate view of everything and then the kind of problem move of data disaggregation becomes more and more meaningful.


So if you look at the majority of data, which is used for city planning for car safety, testing the majority of the eyes aggregated. They threw it across the country to some other people’s opinions sex is not normally distributed evenly distributed which means that are opposed to 50% of females the wild Spa logical fact, so if that’s the case and we using this aggregated day, so which is predominantly focused on men to design cars design cities excetera, then obviously some of the some of the some of the stuff we learned from that is going to leak into pool decision-making Pool City Planning pool

The car designing Industries quite funny because it suggests that women don’t wear their seatbelts property because this motor and have a different builds two men so little bit about the condo importance of data disaggregation and some of the challenges that you face when you go into that later or found a couple of layers and Sons of granularity and get closer to the condo real representative data that is clearly kind of representative of the global reality nonsense without the aid of some slides friend.

If you do me a favor, please push us into some slides and we’ll get kicked off his

By the way for anyone the south at the Disco something they want to say he want to contribute help me out, please who probably tell from my kind of erratic speeds patent.

This isn’t something that I find particularly easy to do without assistance, which is why I use the slides last year.

We did a lot of these strings where we had kind of people joining us. So it was a kind of a two people in the Stream of the same time. So, please feel free to come and help me out. If you’ve got anything that you say know that you understand what you read into this.

Please feel free. I encourage you to contribute and become part of the community service or we can all start looking at these things together rather than me kind of lecturing of people because it’s not really my style and I’m not very good at it. The next one just like it is just a kind of casual for us to come and get on the same page first thing that we want to make clear. Is that just like Real life and the friendships that we fool them recruiting is a subjective experience their little people who try to make it more objective.

Now, the psychometric assessment should exist out that the IQ tests they contribute to was trying to make the selection process a little bit more objective GK something often know the time but more often than the thing that I might swing against can be very subjective. So even if you’re getting really really at your Insight since when individual that’s not to say that the Benchmark comparing against is necessarily as objective is the test that you are being exhibited to

the next one, please bud.

So what is the date to tell us the allows us to understand that there is something going to intrinsically wrong with the way the approaching approaching Recruitment and talent selection. There were a few things that I send confetti Goods in the cases that that might be the case. The number one thing for me is the attrition rates in the UK as we go through the presentation. I’m going to try and sort the data sources.

If I don’t start them right to the point you in the direction of where you might be able to find some of the information in each of the slides are citations so that you can go and investigate this stuff yourself and hopefully because this is really the objective of these LinkedIn live stream.

Hopefully, we give you enough to encourage you and make you feel confident to go out and start interrogating and looking at some of

this information for yourselves and then you find something which you

find valuable and then Encourage you hopefully to go to father and grandfather and get a

little bit more exploratory & Sons of how you do your kind of problem solving and your test reset for me in terms of a marker indicates that suggested that Republicans of recruitment is the attrition rates in the UK according to the office of national statistics 29% of people in

2018 left a job within 12 months.

29% of people left a job within 12 months. So if that’s not really considered as an indicator that something was wrong in the selection process when it’s such a large percentage in 2018 the left within 12 months of that about 20% So you didn’t find a job in the second year.

So it’s not just that they left the first job but maybe moved into another one of the people that was outside for the office for National statistics in the UK 6% very very high ratio of people who are exiting the first joke and not finding a job in the second jet across the

board in the u.s. Productivity is considered to be under or improvements productivity is considered to be under a 2% gain over the last 10 years. And if you consider the things that have happened in Lost 10 years from 2010 to 2020.

We’ve seen the growth of the gig economy which come out and claims to be productivity tools.

We’ve seen much more usage of the internet think I covered it last week there Sunday night full billion people who are currently having access to the internet. From nothing else from 2005 through turn out, right?

So with all of those things, but she’s supposed to be helping to improve productivity was still seeing these real issues and sounds of the growth of productivity in the workplace sent me in the US Senate the US isn’t and indicate what kind of performance after the performance of the global market so I don’t know what is and then finally the engagement rates.

So if you look at the engagement rates the report by Callas in the US you looking at about 32% of the workforce that report continually being a disengaged. I think this one was with with that place now focus on things that point towards an issue with how we were engaging without teams have a building our team’s recruiting people in swat teams. Moritz what else could we be looking at and are encouraged anyone that’s out there that has any ideas to share those so is that we can start taking insulin because of Home story in the office recently with the team where we kind of look up one thing and isolation and then we take a step back and look at it from a slightly different perspective and then step further back until eventually we can begin to understand and I didn’t we’ll decide what we’re really looking at and I worry that sometimes I might be looking at things a little bit too close to them.

So if anyone else got any ideas about how we might stop building more than two sets of broadening at Isis as soon as we get a clearer view of the picture of welcome you to join the conversation.


So that’s kind of the World Views does the indicators of success are some issues, right?

In the selection process one of the things that I feel is a conversation of the diversity and inclusion piece, which is a normal conversation has to be had their significant problems with how people are selected or people are employed in two companies, especially when

they come from a different ethnic group to the people who are making

the decisions


So if we look to pick on Facebook here is just one fairly familiar with from previous experience the same story is vaguely true across all of the Big 4 lacrosse Google Apple Amazon and Facebook is ruled survival representation of males massive over representation of what ethnicity is seems to me like the population is fairly.

Underwhelmingly distributed certainly not evenly distributed the guys at LinkedIn and felt this amazing thing. So they’ve been working on these tools for some time and they have this amazing to local talent insights and it allows you to get very deep into some of the information around Conover migratory what place patterns of Education places of Education the skills the people of recruiting.

Great source for this data by using the information with Facebook make publicly available which to the credit bureaus afford to make Stardust and inclusion days are available through that websites.

Headcount at Facebook and then if you look at the Facebook date, so you can see that that was a rough is 7% increase in the female population in the technical team. And obviously, I mean, I’m not working for Facebook and I don’t want to pick on them even though we all doing a little face but I stopped because as I said, I’m close to it that the relationship in across the agendas and in the headcount of the oval team I feel could do is balancing some alright from my experience.

I mean, they’re not seen his kind of a 50/50 gender split them what happens if you go a little bit further what happens in thedistribution dates are a little bit so we can see a clear or


Question of what the true situation is?

So now we’re going to look at the technical team the business and sales team the senior leadership team and the corner of macro view of the organization the sacrifices by ethnicity and disaggregated by gender split.

So the key story as far as I can see is the senior leadership at Facebook tends to be a very white man’s game. I don’t know it take from that what you well I’m not saying that, you know intrinsically racist of prejudicial, but from from what I’m seeing there is a flavor rolls.

This is my own interpretation of it.

So they they aggregate a little that they it together and they try these statistics about how that performing on that dni stops when you decide different picture. So I think that the message that I see here is the singer in leadership roles of Facebook then if you go and look at the business and sales teams, you can see that I see the gender split.

This is an inference.

This is me looking at it in interpreting it through my own lens, but it does feel a little bit. rolls to kind of different groups different flavors inclusion groups make the girl figures look a little bit more even


This one’s an interesting one as well as I don’t have a look at LinkedIn Talent insights will give you a whole wealth of information. So what we wanted to do after we looked into the schools that were feeding Facebook was look at the diversity and inclusion basa at the school because schools releasing at Publix so of Facebook’s high as last year, it came from just ten schools, but that’s over 4,000 people got Facebook 10 schools only ten schools. And when you look at see the Pick-4 so I can Google Amazon Apple Facebook it something like 60,000 people. Something like 60,000 people came from just 15 schools from the same 15 schools.

I don’t know who I don’t know for me that feels a little bit lazy. I’m not suggesting this intentionally kind of prejudicial against people who didn’t compromise 15 school, but it feels like when they’re done is they’ve identified the feed the schools that have historically for Juiced great candidates for the great employees who who share the the organizational behavior that they want to see the less less likely to display behavior that looks counterproductive. They tend to have the skill-set they speak the same language to they need them to speak and they just couldn’t get on with it fairly quick. See if that kind of fruit.

Which that their arguments for and against it worries me from the perspective of Facebook is when you have full billion fit in a 2 billion people to people using Facebook and the majority of the people who are developing that platform will come from the same school the same kind of social demography Michael similar socio-demographic backgrounds.

What are the the mine says that they have been leaked into the product of so many people using and that applies across Google Facebook Amazon and apple I wasn’t a little bit different because I’m employed so many people come to retail level as well in operational level same as I

will come to that in a minute the statistics statistics from the feeder schools was not terrible at the Ethnicity ethnicity in the school.

Where is Hispanic or other?

I don’t know exactly what lives in although in fact filous inside two or more if you in those car goes in those groups in the knees, but you still went since I went to these 10 schools.


Hey Martin, totally agree with you recruiting is extremely fast, even though no one tries to make his objective as possible compare notes and reflection with all those Etc. But my experience from recruiting over 500 gallons to my company’s is that I often go wrong with skills.

The quite often spawn with culture fit. The culture is to some extent the sum of subjective experiences.

What’s your experience is the bias more geared towards skills rather than culture fit bias is sometimes a little more Insidious than a little less. a little less obvious, so Sick of these on.

So that might be drinking coffee or a specific brand brand Affinity the potential for us to lean into those candidates more than other equally capable candidates is a little bit greater simile.

If you went to the same school spices can can also kick in I think that is less less Define to one specific kind of cats agree whether it be skills.

Willow is easiest to justify it because it’s a well-defined right? So there’s this thing called the Criterion problem with the ultimate Criterion problem, which is why I did in order to be able to

successfully protect the thing you need to have all of the date sources available.

What are the information is available from sleeping patterns to travel plans to diet. I was just simply no possible feasible for us to get over there. And so it becomes very easy to justify the coat to fit peace will justify decision. Is that a pool fit on culture or behavior on the skill side?

Because it’s it’s a checkbox effectively. It becomes a little bit more objective measurable. So it’s hard for me to say that it really comes from one place or another because I think it’s more intrinsic than I think it’s small, deeply rooted in the body in the brain actually coming to sit with us and we have made significant efforts to try and diversify not just on paper, but it was so like a cognitive level or mental level as well.

So we have people in the team who That we’ve we’ve chosen to hire people into the team who they say. They have a different ones that right so that they’ll come in and Conway was the rest of the team might be a little bit quieter. Will they might be slower to make a decision or Foster to make a decision?

Because those different day is a second experiences of those people go through as well. Those are the things that allowed us to widen our lands and widen our view of the thing that we’re doing so is that we can identify potential problem.

We can identify potential Solutions and they don’t cats hiss broadside out of the blue without expecting his time up to see that still going to happen for me one of the key values and having that kind of drug bust in inclusion in the team is that they don’t look at the you don’t

look at the program through the same lens or the time you have different people bringing different considered. To the table. That’s why I think it’s important.

So we talked about Facebook’s hiring passons and the blue my mind when I saw that they were recruiting all these people from the same ten schools, like early this month to finish to put car by Professor Scott Galloway. Tops the kind of financial markets. And so with that. Kind of college my talking about in the thing. I was looking at Facebook and then led me to come to start looking at the guy.

I was interested to see like where to find actually they share the same feeder schools. So the majority of people The large majority 15% I think it was 15% of the people who died by the the big fool the fool came from the same 15 schools. 15% from the same 15 schools and that seems to me to be something. is going to restrict us from necessarily these companies from necessarily looking at the broader kind of social issues of the world of problems that might exist and I felt like sharing these places that are considered as some of the greatest places to walk in the world of recruiting a lot of time from the same places to the next one is from New Hydro function distribution of the 405 percent of the team the Amazon to find in the last 12 months came from just bought schools, but then 17% Facebook came from 10 schools. What was the condom was there ever a reason full that there was outside of my purview.

So I started investigating and segmenting down to engineering and non engineering rolls. That seems to me it’s kind of make a lot of sense right that maybe Amazon is recruit Amazon is what force is much larger is 250 to 300 thousand people or maybe that’s the reason the whistling fumes and tration come from from Phila schools Amazon. The kind of roles of that trying to fill or is Google and Facebook slightly less sign.

Some other interesting stuff that you can get from LinkedIn Town inside. So I think could be quite interesting for people who would not aware of it is migration patents what place migration patterns are quite interesting and just to highlight a couple of them. If you are interested in tracking any kind of company and understanding kind of world it strategy might look like then there’s some stuff you can take from this. So the first thing I noticed was that until he lost a huge number of people’s Apple huge huge number. So in the 1100 people and then I kind of put it in front of some Finance people as I did, you know about this blah blah blah and around and said actually lost year was the end of the Apple acquires the modem business of Intel Core 2 billion dollars. So it kind of made me start to think I wonder if there are any other insights that you can pull from this webpage migration stuff. What place migrations I said that might be able to help us to understand what this costume with a plan of these companies looks like that makes a lot of sense especially to people who specializing in affective Computing the of migrating from a w s y efectiva oil from Microsoft to Google then it help it can paint a picture of what the intentions of those companies off.

Those are things that I feel are important to be in in the condo in mind as we going about using these tools these and these companies don’t sat dates day activity until I feel Google and Microsoft critical pieces of Hamilton them. So understanding what that what that doing what I planned them was at trying to do what they are moving towards is important. It is important for us to know what’s in store for us. and then all of this stuff, thinking how meritocratic is meritocracy anyway, because a lot of these companies Facebook like they promote themselves as being meritocratic the people that one that really deserve and on it and I have a hard time with this because I believe the law of success is a consequence of lock.

I think it’s unrealistic to suggest that you know, just because you’re extremely talented means that you were going to do extremely well. So I started looking into some of the information that’s out there about meritocracy weather will know something that I really want. In my life and I want to say that I think that everyone should be given an equal charms, but it certainly to say that the concepts of marriage isn’t as a utopian as we might believe is the stuff that happens lower down the phone a lot when you were younger. So when you look at that a lot of a lot of your meritocratic O Fortuna see is a consequence of the education that you went through when you were younger and you are absolutely the places that you can go out. There is a lot of data to suggest that geography has a huge impact on Alpha Mobility Plus jumping

Anyone got any questions, please. Feel free to share them. I will do my best to answer them.

So what some of the producer of meritocracy number one measuring and quantifying everything is impossible that is going back to what I was talking about earlier with the Ultimate Party reinforce.

The idea of meritocracy requires that we understand. What are the Merit and SLI of the person that we are selecting are not selecting. There were a huge number of external influences that all And her neon meritocratic. The only way for us to really be able to move towards meritocracy is to discount them for my decision making and then the Lost thing which which was something I read from Reid Hoffman. He he was soaking always talking a lot recently lost two years about

Network being more important than Merit more frequently than not for me is is kind of wave or one of the big influence over even kind of success if you can choose a success as being a capital enrichment well, and that’s what. Write its own network. It seems like network unlock the more important for your chances of success then. Marisa get right being smarter Foster working harder. Don’t necessarily equate with your probability of succeeding. We’ve been there between those things so I have a little bit and also keishin with somebody on LinkedIn a couple of weeks ago and it was really is really weird experience for me Fox as a couple of people who I kind of go on that radar who have some good looking at some of the problems and I said this this research paper with me that was by the stocks and institutes.

When you what is important for the success of what has contributed towards the success the majority of them believe that it was a consequence of that was very interesting to me is a very few of them to be used to this kind of things to do with the class or anything to do with any kind of lock nothing to do with that look gender against another thing. I found last week that really blew my mind and I wanted to share with you that how can how can we be? So, how can we fool ourselves so easily into believing that we we just work hard and we go to the average woman in in Africa was 15 hours completely on paid. And yet the 1% believe that they go there because they work harder than everyone else. What load of nonsense. Frustrates me is one of the guiding principles of outrage by and try and produce some ideas some framework some Concepts and some technology that can begin to address some of that stuff. So if you’ve got any questions, please feel free to reach out to me to Fran.

What’s the cycle my co-founder we be more than happy to take any of your questions to change that so is that we can become more productive as a shareholder value provided to Consumers.

So short-tempered seasoning pool, which is correlated to engagement. Fruit with the wrong person in the first place and its third reason it is not as is because in the quality is not cool.

What is so cool? So kind of coming towards the end of this the the don’t stand it would

be to have a sister. Stay to the inside looking at the days to come together.As one thing. We can break it into components. We can say OK overhead. We’ve got the gender of a heavy coat. Been described previously is the carnival Atomic level of the information.

So is the we can get him up clear review and this is one of the things that I’ve been trying to build a story around in town around if if I’m looking at something that’s closed. All I can see is something that’s close in this I move it further and further away.

It gives me more data are gives me more information through my eyes into my brain to be able to understand what I’m really looking at. It was really happening here. So if I have a if I see a picture of just the coffee cup holder is a coffee cup. I can’t tell you where it is a concert with is in the camp fire if it’s in the house if it’s in a social, what is the really looking at about the story behind that takes rate is On the flip side of that so with multi succumbs more visibility on the

flip side. It becomes much harder to manage. It becomes more expensive to store becomes more expensive to compute and costs a lot more because it’s you go to clean it. Like they seem very rarely comes in a way where it’s just instantly ingestible, right?

You have to go through a process is one of our guys want to buy new guys has recently Discovery. You have to go through a process to get it into a form that you can use took us a long time to get it, disaggregated then into the full might do you need to use it as effectively as possible? So you go these two things and obviously like with cloud computing and Muslim the storage improvements that happening is becoming cheaper and cheaper and she puts his stool and computer hold of this information. The algorithms are getting better. So we’re getting Foster a using it but it is still a problem on the acquisition and the cleaning side.

And then at least listen to something else from Lex fridman. Lex Fridman is a guy who runs a pod cost among other things called the artificial intelligence podcast. He was talking with a professor about a guy walking on around Fitness algorithmic fan and that coupled with another book. So I was reading towards the end of lost chair code for the age of surveillance capitalism as well as some of the white with your hair as well because people often ask us if we use social media feeds and some of my predictive algorithms and I’ll answer is no way I can get into and all the time but the next thing I kind of really wanted to share with you is the idea that in order to get this kind of predictive accuracy or improved in predictive accuracy. We need more Data.

I think the one of the things that we really interested in doing in the next couple years is trying to provide people that use the platform with some controls and measures so they can say OK uncomfortable with sharing this information with you write this Mission.

I’m okay with sharing and then we can then use the information that you’re sharing the willing to show this that you are offering. So it’s made these predictions and explain to the other people on the other side of how actual thing that prediction is a why we can only go

as deep as we can go with it to this. Start to finish before I go to take a break. Number one thing if you have to take away largest three things from today, number one, the chance is all they are hiring patents are telling you a story. Do you know where of is on you to not just learn ways because you don’t just have to learn it.

So I need to discover ways of cutting so you can identify some of the patents it is telling you cuz it could tell you which schools you should you should be recruiting from it could tell you that you’re doing terribly auntie and I fear a large organization that you don’t get that kind of visibility on could even a few ways of suggesting how you might how you might reverse though how you might try to grow that grow

Training a machine with by historical mistakes will only give you more of the same. So what really means? Rather than has come to your organization, so one of the things that we done to is comfortable going on vacation and use your existing teams to give us the kind of known that we’re aiming for right we don’t want to to build a picture of what you have in place at the

moment because if you have had a Issue, that’s cool with you or your recruitment process historically and then we give you the information into the machine. The only thing that you’re going to get out of that is more of the same candidates of hiring so for us we’re really interested in the individual and then kind of objectivity matching that individual back to some stuff like a briefing the machine in the machine doesn’t influence brief.

And then that allows for it with electronic bill. Which is really kind of a job recruitment platform. Don’t just because someone was All Saints. Doesn’t make me in the right person for the job season saline into those characteristics and feel like they are somehow meaningful. I mean, I’m interviewing a Guy that I’m interviewing. You got the moment who wrote a research paper on Hurricane Harvey and yeah, I’m definitely creating some kind of relationship. You shouldn’t doing the try to pull myself. I think we’ve been experimenting this year hopes that are interesting of hope that providing some value and some kind of inside building courage you to go away and think about how you’re approaching your problems alternative ways to eat my approach them and some of the interesting kind of Options in an opportunities that lie inside of the dates are we

Leave a Reply

Close Bitnami banner