Adzuna Australia and New Zealand CEO Raife Watson chats to Simon Townsend, technologist and Principal Consultant from Twelve2, about AI, automation, and what it all means for talent sourcing.
Raife Watson: Thanks for joining our session today. So, I’m Raife Watson, the CEO of Adzuna Australia and New Zealand. And for those of you who don’t know, Adzuna operates in 16 countries, and uniquely here in Australia and New Zealand, we’re actually in partnership with Fairfax, which enables us to access passive and active candidates across more than 150 websites.
Raife Watson: Now why we’re here today, is because when we’re out in the market and such, talking to people like yourselves, we’re constantly hearing all about AI and automation and how it’s affecting jobs.
Raife Watson: And the answer is yes, we do, because we’re talking to many people like yourselves. But, to help us today, we thought we’d enlist the expertise of a subject matter expert, Simon Townsend from Twelve2, the principal and technologist there. Welcome, Simon.
Simon Townsend : Thanks.
Raife Watson: Can you tell us a little bit about yourself?
Simon Townsend : Yeah, so, I’ve got a background in both recruitment and technology. And until recently I’ve been … well, it’s been the last 10 years working in innovation and innovation management, which meant not only helping people come up with and implement new ideas, particularly in HR and recruitment, but it was also around assessing new technology and keeping an eye on the market.
Simon Townsend : I was actually fortunate enough to work with Talent Tech Labs, who are one of the only incubators solely focused on HR tech, based out of New York and I was one of their assessors, looking at the latest and greatest in AI.
Simon Townsend : So it’s a little bit of a sweet spot for me, I guess.
Raife Watson: Great. So the way we’re gonna tackle it today, is just touch on three topics, and do it by way of little case studies. So we figured we’d start … oh, well, the three topics will be, I’ll put them up now.
Raife Watson: How AI automation can affect diversity in your talent sourcing, and how the actual technology itself can lead to errors as well that you need to really watch out for. You may be cutting out candidates you need to get hold of. And third, the actual candidate experience itself. How adopting AI automation into your processes can actually alienate some of these candidates.
Raife Watson: So, let’s touch on the first subject, which is diversity. Now, the case study I want to mention is, look I’ve spoken to somebody we won’t talk about right now, but look they used AI to find candidates. And we all know how that can work. That you can program in the top traits of somebody. So, in this example was, I think, somebody that had got top marks at school, went to Sydney University, studied engineering, also got top grades there, went into their first job in structural engineering and after two years got a promotion and so on.
Raife Watson: So they programmed these traits into the bot, okay, which can also be used to search into your own database, but of course in real time go out across the web crawling whether it be meet ups or personal websites, etc, looking for these people with those common traits.
Raife Watson: But, we all believe, or we know I think now, that this can lead to some problems in diversity. Simon, what’s your thoughts on the pros and cons of using this method?
Simon Townsend : Absolutely. So, anything that matches same to same and like to like, you’re instantly, instantly in trouble in many ways, because do we want everyone in the organization to be the same, for a start? It’s that simple. It’s been proven time and again that through diversity you get innovation. You get change. You get new ideas coming in, and people work better together.
Simon Townsend : But, even so, if you do want a predominantly similar workforce, you’re gonna end up not only with potentially gender and racial issues, socio-economic issues where you’ve got just the norms. Everyone’s the same. Everyone’s the same across the board. You create stagnation. You’re not necessarily getting the best candidate or the best person out there. You’re getting someone who’s performed well in the past. Doesn’t mean to say that there isn’t someone else out there who would perform better. Being able to map traits, map backgrounds, all that sort of thing is, it’s really interesting to be able to say, “There are 30 people out there who fit exactly this profile,” and absolutely talk to them. But, I would be very hesitant to use it as the sole basis of a search.
Raife Watson: I guess the beauty of the technology … it can present those problems but by tweaking it in real time you can see that candidate pool, I guess, grow or shrink, right? So you should be tempted to make sure you put in different traits of these people, I think, perhaps expanding it out. Same same is not always best best.
Simon Townsend : Absolutely not. As long as people are willing to actually play around and change the criteria and see what that does to the talent pool. That’s interesting and that makes a difference.
Simon Townsend : I guess one of the major concerns I have around diversity and any AI or machine learning platform, is that we don’t know what inherent biases the programmer or the developers had when they were creating the software. So there may be … it’s not even so much ones that are within your organization or the way you’re looking for people. The strings you put in. But actually, right down in the code. And that code, we can’t get to. You can’t look at. These are the algorithms and they’re protected. That’s their IP. But how do you know it’s not cutting out people based on their name, or cutting out people based on where they were born … unlimited number of factors that could be an issue.
Simon Townsend : And you never know. Unless you actually get in and understand it. And who’s got a PhD in machine learning to actually understand this stuff? That’s something that you really need to look for.
Raife Watson: Okay, so I guess what we’re saying here is that, inherently in the code itself there can be a problem. So, you should, if you’re gonna use this type of software, make sure you cast your net a little bit wider.
Simon Townsend : Yeah, I think …
Raife Watson: Secondly-
Simon Townsend : Sorry, just on that. Using multiple pieces is probably a good thing. Because the other thing is, any … so I’ve been doing a bit of, sort of, executive head hunting recently, just to keep my hand in and keep assessing the various tools. And I’ve found that if you run, I’m not gonna name names, but running sourcing tools to identify people was identifying a pool of, let’s say, 50 people that will meet the criteria. If I went out and did the old school Boolean sourcing and the various other techniques that I used to use as an online sourcer, I was finding a pool where there was say, 50 people, and there was a crossover of about 15, 20. You know, it’s not necessarily … you’re searching in different pools. You’re searching in different areas. So an inherent bias in the software could be just the sources that it’s searching. So if it’s only searching through LinkedIn, you’re only gonna get people who are in that space. If it’s only searching through Twitter, etc. etc.
Simon Townsend : You’re only going to get what the developers think was important.
Raife Watson: Is it cost-prohibitive to use multiple pieces of software for sourcing like that?
Simon Townsend : That scale? Yes. I would imagine so for many people. Then again, it depends if you can, you know, potentially reduce head count that there’s a large cost reduction there, potentially.
Simon Townsend : But, most of them offer free trials, free test periods, at the very least they’ve got the information on the website and you can go out to the various forums and sources and like Facebook and that sort of thing. Talk to the community who are using it, and get an honest assessment from them, the actual end user. Incredibly important.
Simon Townsend : So, it doesn’t matter what you’re using. If you ask about it online in the right groups, there’s gonna be a bunch of experts that are gonna tell you exactly what their use cases have been. How it’s changed their recruitment processes or their sourcing processes.
Raife Watson: Right, so I guess the message is, cast that net wide yourselves before you jump into that software. And that brings us to, I guess, the second case study or piece of topic is, how using this technology can lead to errors. So, what we’ll use here … sorry, just skipped one slide. What we’ll use here is example of our own software. We have a team of data scientists that build modes for us, and one of our products is ValueMyCV, which allows the candidate to come onto the site, upload their CV, and then the model basically tells them how much they’re worth in the market, suggests other jobs based upon their CV, corrects mistakes, etc. etc.
Raife Watson: But to do that we’ve gotta use parsing technology at the top of the tree, right? So we’re reaching in and parsing that technology in to the model, as such. However, let you in on a little secret I guess, is that we’ve had about 100,000 CVs loaded in the last, since the beginning of the year, but 5,000 of those instantly get dropped through this parsing technology. So, 5% maybe doesn’t sound like a lot, but I think we all know, well that’s 5,000 CVs right there out of 100,000, but we know it can just take one good candidate, especially in a candidate-scarce market, to make all the difference. So, what’s your opinion on the use of that technology, and how can we avoid, maybe, some of those pitfalls? And what have you seen in the market yourself?
Simon Townsend : That’s another big issue. And just on RateMyCV, I was actually really annoyed because you got my value almost bang-on, and I’m unique. I’m special. I innovate.
Raife Watson: Aren’t we all?
Simon Townsend : It was within $2,000. Right? Okay. It was really nice.
Simon Townsend : So, I’ve got a friend in particular, the case that I’m thinking of. He’s in innovation, and his CV is amazing. It’s infographics. It’s … there’s stories, there’s journey maps, there’s all the things that make him an amazing innovation person, and it’s got the stories of how he’s helped grow multiple start ups from zero to exit. His CV is not gonna get parsed by any of the ATS’s that I know. Because it’s pictorial. Because it’s spread about in weird places. He has no chance. He’s a fantastic candidate, but he will never be seen by the systems.
Raife Watson: So that means you need to put a bog standard CV in, essentially, because you’re wary of it throwing out if you put a little bit of coding in there or something to show your work or, I don’t know, something like this.
Simon Townsend : Yeah, that’s actually what I ended up advising him to do. Put at the end of it, here’s just your old-school standard CV. And it’s not just, obviously, parsing technology of CVs. It’s all the way along the way. If there’s one bit of error that makes the system scream, “I don’t get whatever that is,” in order to stop it breaking, the software developers have to put in, “Well, if error, move to whatever and ignore.”
Simon Townsend : So whether that’s, again, it’s something that pulls out and is looking at your Klout scores combined with your DiSC profile or whatever it might be in an automated fashion, starts matching it in and says, “Actually, this person doesn’t have a Klout score. What do I do then?” It could essentially drop the entire person. So you’re losing an entire person along the way.
Raife Watson: Okay, I’m trying to think on the fly here to be smart. So maybe, with candidate-scarce high-level, or maybe like data scientists, right? Because that’s the rockstar of the industry at the moment, right? So there’s only a few of those around, really. So maybe for a process like that, you’d put them outside of the process because you’d only expect a minimal amount of CVs to come through, and you don’t want to lose them. [inaudible 00:12:03] possible solution, do you think? Is it old-style [crosstalk 00:12:06]
Simon Townsend : That’s an interesting thought. I like that. And it’s almost putting executive search-level techniques onto the roles that are more important. Not necessarily at the executive-level, so you’re doing a more thorough search where you’re using a multitude [inaudible 00:12:19] plus humans. Artificial Intelligence isn’t there yet, and I used to be bothered by the difference between machine learning and artificial intelligence and I’m trying to get over it. But augmented intelligence is the key, I think, right now. So not necessarily trusting that the software does everything perfectly, because it won’t. It’s software. It will break or it will go down. But being able to wrap that into your best practices. Identifying who those key hires are, and putting special kid gloves on for them. Maybe you’ve got 100 hires a year, you’ve only got three data scientists that you’re looking at. Focus on those and let the systems deal with the other volumes that you’ve got.
Raife Watson: Volume recruiting, yeah.
Simon Townsend : So, you can be smart about how you use it, and it’s not that those people are any less important, but they’re potentially less important for your business at that time. So understanding the reward and the risk, and where you need to throw the human element to really ensure that you’ve found all the candidates, and that they’re being treated properly, and they’re not having a bad candidate experience because they’re being thrown out or just ignored.
Raife Watson: I think that probably brings us to the next piece, right, which is how can we … how’s the candidate experience affected by AI? Now I know that we’re all in a bit of a rush to implement some of this technology. We don’t wanna get left behind, and it seems to be everywhere. But how’s that gonna effect the actual candidate experience? I mean, we sort of touched on it here, I guess, but I don’t know if anybody saw the news yesterday, I think it was, with Telstra’s chat bot Codi, it was driving people crazy, I guess, and I think it did a lot of damage to Telstra’s brand over the last couple of days where the chat bot, I think was not answering the questions, or answering all different questions the same way, etc. and nobody could get hold of an actual person to discuss this. So, it hit a lot of chatrooms yesterday all about that.
Raife Watson: So, we’ve gotta think that if you’re implementing this type of software, and there’s a popular brand, I guess, or technology solution and you’re implementing the same one, are we now commoditizing the process, you know, especially with these rockstar data scientists whoever they could be that you’re looking to differentiate yourselves, can you, if you’re using auto-scheduling and chat bots to answer simple questions, etc. and does the candidate not feel special and then deviate towards, maybe, good old human touch speaking to a recruiter process, which could lock out some of the larger firms, or larger recruitment agencies, possibly. Thoughts on that on the candidate process?
Simon Townsend : Absolutely. The commoditization of the candidate experience is a really interesting line. I like that, as a sound bite. I’m gonna steal it off you.
Simon Townsend : So if you look at chat bots in particular. Top line around candidate experience. Some people love them, some people hate them. So there’s an interesting outfit, company out of Hong Kong called Talkpush who do … will create a chat bot for you in Messenger, Facebook, wherever you need it to be. Could be in Slack. Doesn’t matter.
Simon Townsend : As you interact with it, it does a little bit of an assessment of you. Says, “Are you interested in X role? If you are, do you wanna talk to me now or later?” If it’s later, it’s got the automatic scheduling, if it’s now it will instantly call you, that sort of thing. And by no means the only one. You need to look at Olivia, you need to look at Mya. There’s a whole bunch of them out there at the moment, it’s just I know Talkpush from when they were dealing with the innovation labs.
Simon Townsend : And the candidate experience for technology people, for people who are younger. A younger demographic with a highly technical background with technical skills. They loved it. They didn’t interact with someone. They got to the point where they were running through, they had the automated phone call where it was a number of questions being asked that had been set by the recruitment team.
Simon Townsend : The answers were recorded, turned into text and sent back, as well as video blocks. It’s an interesting platform. At the end of the phone call, 80% of the people that they were targeting in that particular demographic were saying, “Did I get the job?” They were happy to be assessed completely without a human in the mix.
Simon Townsend : That said, you see other things, and … take video interview. So, a one-way video interview where you’re being recorded as you answer questions. Again, it completely divides people on a candidate experience. Some people absolutely despite it. Some people love it. It’s okay, they can talk to the camera. They’re happy to do it time-shifted. So, I think a lot of it depends on the roles and the pieces that you … where they fit, the kind of demographic you’re searching for.
Raife Watson: We’re talking about age, right?
Simon Townsend : So, you see, it becomes very difficult.
Raife Watson: We’re talking about mature-age workers now, and that means from age of 46 and over, right? So some of these may not be so technologically savvy or wanting to interact with technology, right? And some of these will be applying for lower-skilled roles, be it retail or maybe they’re actually senior roles because they’re older now and have reached these positions and they don’t want to or cannot do this.
Raife Watson: So, again …
Simon Townsend : Potentially you’re creating a huge hurdle for people that could be seen as a bias to the point where it’s actually illegal. I mean, theoretically.
Raife Watson: Because you’re locking out a segment-
Simon Townsend : Absolutely. You’re completely ditching a segment of people and saying, “I’m not providing you an experience that is conducive to you getting the job.”
Raife Watson: Which loops back to diversity, right?
Simon Townsend : Yeah, absolutely.
Raife Watson: [crosstalk 00:18:38] I feel like [crosstalk 00:18:39] anybody who can do this or is that same type of talent.
Simon Townsend : Yeah. The other element around the candidate experience is the actual interaction itself. So, where you’ve got things like Codi throwing out the strange answers. That’s okay, and it’s funny to a point, and people are willing to accept, okay we’re trying something new. It didn’t necessarily work.
Simon Townsend : But not being able to get someone on the phone to say, “Give me the actual answer”, that’s a huge issue.
Raife Watson: Yeah, have an option, right? So, damage your EVP I suppose, or just damage your brand. Damage your hiring.
Simon Townsend : Absolutely. And again, back to the commoditization of the experience. I find that particularly fascinating as a line, because I’ve, you know, built a couple of chat bots a long time for various people, and it’s hard. You have to have a lot of information that’s pertinent to those people that you can train this chat bots with that you can make sure it understands what it’s doing. But the basic hello, how you doing, what’s going on, set of responses, you can pick those up offline, but it’s commoditized. It’s the same experience that people are gonna get on those chat bots.
Raife Watson: Are there any good examples that you could send people to, to have a look at?
Simon Townsend : I would definitely be looking at the IBM chat bots. They’re very quick and easy to get up and running. There’s … it’ll come back to me in a second. There’s a particularly useful one … name escapes me. Sorry. I’ll come back and I’ll pass you some information about that. But, yeah, there are a few in particular that are … if you’re just looking for what is effectively a customer service bot, so that’s something that people have spent a lot of money on. So if you just want basic Q and A, here’s point to our website, here’s a bit of information, whatever it might be. They’re not too hard to set up. It’s not easy, but you can do that.
Simon Townsend : If you’re looking for a full end-to-end platform, then I would definitely be looking at Olivia. I would definitely be looking at Mya I would definitely be looking at Talkpush, and there’s a few others as well. They’re getting close to the point where it’s not necessarily removing a recruiter, but it’s removing the heavy lifting across a lot of the talent acquisition platform.
Simon Townsend : And that’s what we’re all after, right? It’s actually how do we embrace automation as it does the skills and the sets of tasks that are either onerous or difficult or repetitive, to enable us, should be, to enable us to talk to the candidate, to provide them a better experience. To give them the right information. To talk to our clients and understand what the hiring manager’s needs really are and provide them with a better situation.
Simon Townsend : Or work on the strategy of it. You know, what’s the next step in the process? How do I develop the next piece? What’s gonna work best for us?
Raife Watson: And technology’s moving so fast. So, would your advice be to anyone out there, to go boots and all, because it’s gonna modify on the fly and that’ll be great? Or do you hold off? How do you test it out?
Simon Townsend : Test and iterate? So, as a design thinking proponent it’s all around identification of the problem. Find a couple of potential solutions. Test them small and iterate. So, if you can throw just one role, or one set of roles if you’ve got repeat roles, through a process where you’ve got automatic sourcing [inaudible 00:22:25], a chat bot to reach out to them and do an initial screen, and that point they come in to talk to a recruiter, or something to assess the CV-
Raife Watson: And always give an option in these to opt-out and go speak to a person.
Simon Townsend : Absolutely. Especially, no absolutely. I was gonna say especially if they’re reaching out to you, and your brand. But if you’re reaching out, it’s the same. Always have an option to, I need a hotline, red button here. Let’s talk to the Kremlin type situation.
Simon Townsend : But try it small. And I think that with the number of developers and platforms that are coming up out there, there’s a lot of people who are willing to let you try and work with you and partner with you. It helps if you’ve got a great brand, Telstra or something like that. Even the smaller enterprises and down into the SMEs, there’s always someone out there willing to partner and work with you. So it may not by Mya, because they’ve got some great traction in the market already, and they’ve got a product worked out. But, if you look at the amount of money that’s being tipped into start ups in the tech recruitment space or the tech HR space. Literally 100s and 100s of millions of dollars in the US in the last quarter alone. So, there’s people out there.
Simon Townsend : Keeping track of them is hard.
Raife Watson: Well, that’s what I’m thinking. Thinking, well, how do you know where to start? That’s what I guess-
Simon Townsend : It’s help to talk to a consultant who talks to those people.
Raife Watson: True. True.
Simon Townsend : It’s understanding that they’re all trying to get their product out there, so there are marketplaces and forums where they’re trying to understand who their market is. Being able to identify someone interesting who isn’t just cut and paste someone else’s work, has something new to bring to the table. That’s interesting, but it comes with a level of risk. So understanding your talent acquisition profile’s risk appetite is important, and that’s where I say, if you could do it with just one role, you’re actually minimizing the risk. And if you do it in conjunction with a standard process, so put that same role through both a new process and your standard process. It’s like A/B split testing. So, you can actually assess them both and understand what works. And you can do it at very, very minimal cost because you’re only putting your toe in the water.
Raife Watson: Well, I guess since we have you here, it’d be great if anyone has any questions, please use your sidebar to give them because Simon’s right here, and he’s online to answer any questions. So, if you can submit them now we’ll have a look, but I think the key messages here are weigh up your options, don’t rush headlong. Maybe speak to a consultant, look at other companies, etc. to work out whether it’s for you. Whether it’s for you right now, and whether it suits your candidate journey or what your sourcing needs are.
Raife Watson: And maybe it’s not a one-size fits all. Maybe it’s works for one particular stream of people you’re looking to hire and not for others. And of course the age and diversity can come into that as well.
Raife Watson: Look, while we’re waiting … we’ll have a look for these questions. Just wanted to also iterate, look, please, we’re both contactable. So, I’m simply email@example.com and if you have any questions on how Adzuna can help you … so, we’re not a one-trick pony as such. We obviously have our site, but we’re able to take your ads and look for that passive candidate. And I think some of this talk about AI is getting out of that traditional active candidate pool that’s just on a job board or job site, and looking for them across other sites. And that’s what we’re able to do.
Raife Watson: Also, displaying your brand across 100 different websites, whether it’s the Fin Review or The Age, etc. We can find these candidates that may not necessarily be looking right now.
Raife Watson: So let’s have a look to see, also if we can find any questions on here.
Raife Watson: Oh, here we go. We’ve got one from Geoff which says, “I’m aware of Mya, but can you repeat the other bots mentioned?”
Simon Townsend : Absolutely. So Olivia, from Paradox Software. They all seem to be female names. I guess that’s because … there’s a whole piece around diversity and how we should be naming things in there. And Talkpush. So, Talkpush based out of Hong Kong. Paradox, and with their bot Olivia, based out of the US, and Mya out of the US as well.
Simon Townsend : There’s also some really interesting stuff coming out of Israel. So, it’s one to keep an eye on, if you’re looking for … and Australia, I should say. There’s lots going on here, but if you want someone local, there’ll be someone out to … [inaudible 00:27:25] one of the incubators, whoever it might be. So there’s always an option as well. You can sort of walk into one of those incubators or accelerators and say, “I’ve got a problem” and there’s 400 entrepreneurs who’ll put their hand up and go, “I can help with your problem. I will solve your problems.”
Simon Townsend : Australia’s got a strong start up community.
Raife Watson: Yeah, yeah. Okay.
Raife Watson: Now, I’ve got another one here, which is, “What the best responses to avoid when building the bot for a business?” I don’t know.
Simon Townsend : Ah, depends on your brand. So, when we developed the Twitter handle for Deloitte, it was … we used the actual branding itself, so it was a round, fat jolly green dot that rolled around and did fun things, initially. I mean, that was our first thought when we were building the social media profile for it.
Simon Townsend : And Deloitte was fun with serious intents. That was the element that we took …
Raife Watson: I like that. Fun with serious intents.
Simon Townsend : So we took that and absolutely fed that into the social media handle. So when we interact with people, it was allowed to be a little bit cheeky, but it had to be very professional. It depends on your brand. If you’ve got … I mean, there’s obviously things, you can’t ask someone’s age necessarily. It depends on the area you’re in. It’s a bit of a mine field.
Simon Townsend : But, assuring that whatever you do is gonna hit your brand right, and keep pushing that message and that branding out there, that’s key to any questions you develop. Would my brand, if I’m, you know, an insurance agency. “Is it okay for me to be asking people this question?” Is the question you ask every time. Which may well be different from a swim school or a childcare center or whatever it might be.
Raife Watson: Okay. I’ve got another one here. I’m not sure if that reads right, but “Can you ask the bot to expose that you are not speaking with a human?” Do you think that means –
Simon Townsend : A lot of the bots introduce themselves as bots these days. There’s been a bit of a kickback against the faking it of, “I’m pretending to be a human.” And it actually also leads to problems where, if you think you’re talking to a human and they start doing something strange, then you’re left with a really weird … it’s like a reality gap. Which is interesting in cartoons if you … anyway. Side talk.
Simon Townsend : So, you can always ask if you’re typing. If you ask, “Are you human?”
Raife Watson: Are you a robot or are you human?
Simon Townsend : Yeah, the vast majority of them will respond with, “No, I’m not. I’m just learning to be a person” or “I live here in this space” or whatever it might be. So people have got canned answers for that.
Raife Watson: And is that where it should say, “Do you want to speak to a human?”
Simon Townsend : Yeah, absolutely.
Raife Watson: Hit bang. Yes.
Simon Townsend : That would be key. It becomes, I guess, harder when we get to the point where we’re doing voice recognition, and we’re combining and old-school IVI with a bot, so you’re talking to someone and you don’t know you’re talking to them, rather than typing to them. Because it’s difficult at that point to say, “Oh, sorry. Are you actually a person?” That’s hard. And that’s coming.
Simon Townsend : But when you’re typing, that’s fine to ask. No one’s gonna be offended on the other end of the line.
Raife Watson: Okay. No. We’ve got another one here to, which is, look, I also can see that some of you have had difficulty hearing, so we will be sending the presentation around. But certainly you can contact both Simon and myself to ask other questions.
Raife Watson: And then I’ve got here, “What are your views on video interviewing?”
Simon Townsend : It depends where it is in the process and what you’re using it for. How you’re using it … video absolutely has a role to play in the recruitment process. I have no doubt about that.
Simon Townsend : I personally don’t enjoy using the platform. That’s just me. I don’t like looking at myself on video. If it’s two-way video, and the stuff’s being recorded, split up and sent to various, I think that’s great. People can see my answers if I’m being interviewed. Various hiring managers can see the answers, I don’t necessarily have to go over the same information multiple times. That’s interesting.
Simon Townsend : Where it gets really interesting is a platform like HireVue, who are doing facial recognition and AI and machine learning algorithms in the background to understand when someone’s not necessarily being truthful, or when they’re uncomfortable or whatever it might be. And I believe it was HireVue, I may be wrong. I believe it was HireVue. They actually sort of flipped that on it’s head and looked at videoing the people who … the hiring manager.
Raife Watson: Right. That’s dangerous.
Simon Townsend : And they were able to then identify inherent bias within them as they were going through, and say that this person is kicking out people who are qualified because they don’t like their gender or their religion or their whatever it might be.
Raife Watson: And is that being used in Australia here? HireVue, I don’t know,
Simon Townsend : HireVue, yeah. The platform’s available. It’s a global platform. Based out of the US, but it’s absolutely a global platform. I think if you’re doing volume for scale, then I would be talking to Sonru. If you’re doing a two-way process then I’d be talking to Montage. If you’re looking for some really interesting, funky stuff, I’d be talking to HireVue.
Raife Watson: Okay, great. I’ve also got another one here. [inaudible 00:33:12] that “Can you suggest any new recruitment system that actively uses AI for assessments?”
Simon Townsend : As a whole system? I’m not sure of any that does it across the entire piece. I mean, you’re really looking at the Myas and the Olivias at that point. So they’re taking the sourcing, the screening, the interviewing and putting the machine learning around that piece, rather than the actual ATS. I’m not sure if I’m answering this question correctly. I’d love to understand exactly what you mean by that.
Simon Townsend : There’s an interesting platform based out of Israel called Comeet.co.
Raife Watson: Comeet.co?
Simon Townsend : Yep. Sort of like comet with an extra E. So Comeet. They don’t necessarily have machine learning over the top, but what they do is they’re changing the process so that everyone’s being brought in from the organization at different times and it’s sort of clever algorithms that’s making that work. So there’s a couple of people playing at it from different angles. I don’t know of anyone who’s actually got just a standard ATS that runs machine learning all the way through it.
Raife Watson: Okay. Now we’ve got time for one more question, then I just wanna talk about our next webinar, which we’ll do better at I think for the sound, which was a question I saw her about Google and Facebook, I think it was.
Simon Townsend : Yes.
Raife Watson: Oh, okay. So did you see that question, I just lost it now. I think it was about, like, how some big start up, whether it be Google or Microsoft, have already built bias in or something. What do you think about that? Or something like that? Sorry, I can’t find it again.
Simon Townsend : Yeah, so that’s where you’ve got the developer, without meaning to, absolutely without meaning to. As they’re creating the code, if they don’t have a true understanding or the experience to understand why they’re cutting people our or why these biases may be creeping in, that’s a huge issue. And that’s my biggest concern right now, is the fact that people pick up something new and shiny, because we’re humans. We like a shiny toy, and if someone says, “You’ve got a little bit of money, go out and play” or “you’ve got a lot of money, go out and play. Find a new thing for it in the process.” You’re gonna go find something new and exciting and shiny.
Simon Townsend : At the heart of that shiny is a black box. And that IP that’s held in that black box, the way the developers have created it and programmed it. If we don’t run that in parallel with other tests to understand where people are being pulled out, why they’re not having a great experience, whatever it may be. If we just trust it and we run with it, that’s potentially really dangerous, because it just feeds the algorithm itself. It starts matching again, and if it matches and it’s not finding people who are born in Scotland or that sort of thing. Whatever it might be. However crazy it could be.
Simon Townsend : Once it starts repeating itself and repeating itself and learning off its model, it’s not gonna get the edges, it’s only gonna get the center, and you’re gonna have that same problem amplified dramatically. And it’s probably not by design, it’s just they don’t know.
Raife Watson: No. Okay. Right, well look. I think we’re basically out of time, so. How do people get hold of you, Simon?
Simon Townsend : Simon@twelve2.net is my email address, or twelve2.net online is where I live.
Raife Watson: Okay, great. And look, we’ve got another webinar coming up, as you can see on the screen, on the 18th of April. We’ve got another guest. Look, I swear that’s not a chat bot. Not Guglielmo. But he’s from … Doctor, in fact, Guglielmo. He’s from the Behavioral Insights Team, which is all about behavioral economics. How little changes in processes can have a big effect, okay? Which, on sourcing, is huge. I think you said you were very keen to see that as well.
Simon Townsend : Yeah, absolutely.
Raife Watson: And that’s all about effecting diversity as well. So that’ll be on April 18. Hope to see you there. Please contact myself, firstname.lastname@example.org, R-A-I-F-E, or Simon, and we’re very happy to chat, and we’ll send out also a copy of this. And apologies for some people with had a sound problem. Thanks again.
Simon Townsend : Thank you very much.