(upbeat music) - I am super concerned about people, how people approach and use AI and data in recruiting or any type of matching. And that matching is at the college level, of admissions, or thinking about opportunities, skillsets, those type of things. And the reason I'm concerned is because our algorithms are still pretty dumb. They don't have the sophistication, it's still early days, and the data sets, technically we should use a technical term, they suck.
Because if you look at who gets into these colleges, or who gets into a certain role, you don't have examples of people who didn't fit the mold. And so it's very overly biased. And we have to be very careful with that because otherwise we say, "Well, who gets to be a VP of Sales? "Oh you have to have an MBA, cause everyone has it," but there's never been a set of people who we've tested who don't have an MBA. So these data sets are biased. And we know this is concretely true because we've seen it where people have tried to use these algorithms and then suddenly the algorithm is doing something that is literally racist.
Because these algorithms are black-boxed, it ignores or rejects a name because it's a name of a particular ethnicity, like specifically the African American names. So, we have to be very, very careful there. There's another part of HR where I think we're going to see a great opportunity around these algorithms and that is to find out when we are ... we have bias. And if the machine's like, hey there's actually, you know all you're doing is hiring people that look the same or a certain ...
Certain one-dimensional, the algorithms say, hey that's not ... That's not really okay. That's like flagging this for you. I think that's the way we'd like to have machine learning and AI start to show up, is saying like hey, maybe you should consider the following. Another dimension and this was launched in a project called LinkedIn Skills, that moved into Career Explorer, and then LinkedIn Endorsements as well, is this idea of where do skills transfer. And this largely the genesis of this project originated watching my friends come home from Iraq and Afghanistan after service, and not having the ability to say, hey how does my job act ...
Or the work I've done translate over. And so, there's this sort of mapping in the world, if you will, of skills. And what we haven't been able to easily do is say you know what this cluster of skills, you may have one of these skills, but you don't realize that your skills actually fit all over here. - [Interviewer] That's very interesting. So it's really using unsupervised machine learning to create these clusters that we never thought some skills would be bundled into. - Right. - [Interviewer] And so you're really defining a brand new space with that. - Right.
You know, there was a joke in the early days of the LinkedIn data science team is we had a neurosurgeon on the team And who had given up neurosurgery 'cause he just loved data and problem solving. And so, the motto of the team in the early days was it's not rocket science or brain surgery, but if it was, we'd still have you covered. (upbeat music)
Skill Level Intermediate
Wrapping up1m 5s
- Mark as unwatched
- Mark all as unwatched
Are you sure you want to mark all the videos in this course as unwatched?
This will not affect your course history, your reports, or your certificates of completion for this course.Cancel
Take notes with your new membership!
Type in the entry box, then click Enter to save your note.
1:30Press on any video thumbnail to jump immediately to the timecode shown.
Notes are saved with you account but can also be exported as plain text, MS Word, PDF, Google Doc, or Evernote.