Name: Pymetrics: The Game-Based AI Recruitment Platform
Uploaded: 2019-12-27
Duration: 5 min 31 s
Description: Here’s what to know about the tech startup that is said to enable blind hiring.

Pretend you're a hiring manager. After an initial glance of these two resumes, who would you hire? [ MUSIC ] Resumes should absolutely not be the main decision factor in the hiring process because they enable structural racism, sexism and classism. And hiring is broken Frida is the CEO of Pi metrics Biometrics uses Behavioral Science to look at things that are more fundamental about people so their cognitive social and emotional aptitudes and we use that data to understand somebody's fit to a job. And we replace traditional methods of evaluating people such as looking at what school you go to or who you know at the company. What happens is you apply to the job in the same way that you normally would, right. You go to a company career's page, you fill up the application and either write there in line Pymetrics popped up and it says great, this is part of filling out your application, you're gonna go through these 12 games, that are gonna help us learn about you on a more personal level. And then you complete the games right there, we're able to then calculate a fit to that role, as well as all of our other roles on the backend and send that back to the client. Or as a candidate, I can submit an application and then get an email that says the next step in your hiring process is to go through pymetric. We have no idea about any of the demographic variables that are very very like front and center in a resume. We know nothing about you. So it enables blind hiring basically all we're saying is like, do you seem to be a good fit for the job that you're applying to? And irrespective of any of those other demographic variables. Here's something similar to a game you might experience from Pymetrics. Every time you see a red circle on the screen clap [ MUSIC ] Permetrics claims a game like this can measure levels of impulsivity, attention span and how you learn from mistakes. Permetrics did not invent these games they actually come directly from neuroscience research exercises. The games measure qualities that are really inherent to the way people behave. So these are things like the way you plan, the way you take risks, the way you pay attention, how you pick up emotion and people, all with the idea that there's actually no good or bad way to be, but that behaving in different ways might make you better for certain jobs. Apps. I'm Inie Shragden, I'm a PhD candidate at Cornell University, studying computer science and in particular bias and discrimination in machine learning. Manish's latest research examines how companies like this use platforms for hiring. What we found with [ UNKNOWN ] in particular, is they're a fairly well- established, large company. They think a lot about how to comply with the law and how to do things in an ethical way. And I think the limitations that they suffer sort of common across the industry and that no one really has a good idea of what best practices look like here. All it's doing is an all any machine learning is really doing is trying to replicate past decisions or replicate past outcomes. And so they're only as good as the data that we give them. And ultimately, if we're feeding them data that we've created through our historically biased hiring processes, they're going to struggle to be significantly better than that, without a lot of intervention and painstaking testing to make sure that they're not incorporating our human biases. There are such things as algorithm bias, yes, we have to be careful, we have to build algorithms and technology in a way that mitigates those biases, but I think if we just throw out AI or we throw out technologies as a solution all together, we're really throwing away the baby with the bath water and we're kind of backing the situation where we gonna try to fix human beings, and again the last 30 years have not been proven that we can do that in any successful way. It's not totally clear how well these particular games that they're using have been validated against the particular jobs that they're hiring for. Everything is very context dependent and the research literature just isn't comprehensive enough to tell us that every game is going to work in every situation. We work with a large number of Fortune 2000 companies across the globe. Clients that I can mention publicly include Accenture, LinkedIn and others. And our clients have seen a huge increase in gender diversity and ethnic diversity. And one of the main reasons that they're interested in using a technology like this is because they are aware of human bias and they want to Mitigate it. And so seeing those increases in all the types of diversity that they're wanting to increase is a really powerful testament to why they're using the technology. The assessments are here to say it's likely that the use of machine learning algorithms is also here to stay. And it's really going to be a question of how do we build, validate debias these tools in ways that we're comfortable with in ways that are transparent. And hopefully in ways that can be regulated by sort of governing body with appropriate strength. So I think if we are very careful about our design of technology, and I think if we implement auditing of that technology which penetrates. [ UNKNOWN ] does at many points in its usage, as well as thinking about technology as needing safeguards or guard rails around it. I think absolutely we can use that technology to improve the process. If every company use plyometrics, I am certain that we would have a more diverse workforce from the perspective of gender, ethnicity and socioeconomic status and I would be various confidence that people would be much happier, more productive and stay longer in their jobs. I have no doubt about that. [ MUSIC ]

Pymetrics: The Game-Based AI Recruitment Platform

Careers

This tech startup is said to enable blind hiring.

Here’s what to know about the tech startup that is said to enable blind hiring. 