AI and Why the Future of Work is Working With People

Andrew McGlinchey

It’s no surprise that Andrew McGlinchey, Senior Director of Product at Indeed, got his first computer at age seven and a degree in AI way before it was on everyone’s radar! Though, he’s come a long way since building his first chatbot as a kid in Canada. He’s now based in Southeast Asia leading the establishment of Indeed’s Singapore Engineering Centre. I sat down with Andrew to pick his brain about hiring for tech talent in tight labour markets, what excites (and frightens) him about the potential of AI, and what the future of work is going to look like.

Everyone is talking about AI, but I’m not certain everyone actually knows what it is. How would you explain AI to a 10-year-old child?

For things that aren’t AI, somebody writes rules; ‘if this, then that’. So, if you’re playing a boardgame you could read the instructions and know what to do, such as if you roll a six, you move six spots on the board.

Compare that to what we usually call AI, which is more like making judgements. So you teach, rather than give rules. You teach a computer how to do something, like determining if something is a picture of a cat. People are very good at this. But to teach a computer how to do it, you need to show it millions of pictures of cats and millions of pictures of things that aren’t cats, then it will slowly work out how to tell the difference.

Do you think AI is contributing to or helping to eliminate bias from the hiring process?

It has the potential to do either and the problem is that bias is super hard to detect. How do you measure the bias? These are difficult decisions anyway. When a hiring manager, without AI, says ‘yes’ to a candidate, or ‘no’ to candidate, how much was it based on bias? I don’t know, they don’t know either. People make assumptions, sometimes implicit, sometimes explicit.

Here’s how AI could help; we train our systems not to look at things such as gender and name, so they’re not looking at things that humans are looking at when we know they could be biased. On the other hand, our training dataset is done by humans, so is there a risk that we’re codifying humans’ bias? Maybe? But maybe other things correlate.

How does Indeed use AI to help people get jobs?

We have trained our systems to do a lot of things. All of the jobs we aggregate are just words and yet when someone types one or two words to search for a job, we can find a shortlist we think they’re going to want to apply to, from the millions we know about.

And how do we do that? Sometimes it’s a principle called a doppelganger. So we look at people who have spent more time on the site than you and the types of jobs have they clicked on. If someone clicks on jobs A, B and C and you clicked on A and B, we can recommend C. That’s one way of doing it, but often we can figure out more general patterns.

So, after we figure out which people are like you, we figure out which jobs are similar to the ones you have clicked on. When it comes to employers we can do the opposite and recommend people to you based on similar principles.

Another example: we have built a machine learning system to predict a salary range for certain jobs based on its description.

Can you envision a future that phases out all human contact in the recruitment process?

Probably not, because the future of working, is working with other people.

So if your job is sitting in a room looking at scans to determine whether there’s a tumour, a machine may be able to do that repetitive part for you. But the people part then becomes even more important. Anything that doesn’t involve working with other people is mostly drudgery work we can often do away with. The software can help us do less of the boring stuff, like filter through many applicants who are clearly applying to the wrong job, and free up more time to conduct quality interviews.

Reflection of a man in a window on his cellphone

What aspects of AI excite you?

It feels like there’s some amazing things were just starting to get at as we tie different pieces together. We’re getting smart at things that used to be impossible, like detecting the sentiment from a Tweet, or identifying a face in the crowd, or recommending songs or articles that you’ll enjoy. It’s amazing.

We’re starting to get ‘adversarial networks’ where there are no ground truths. Two systems working to beat each other, such as one that says can I draw a bird, and another that says can I detect a bird. Together they make things that are pictures of birds that didn’t previously exist.

Professor Toby Walsh advocates limits to AI so it’s used to enhance not damage our lives, are there any use cases for AI that frighten you?

Part of the problem is we can’t double check it, we don’t know if it has used some form of bias that we didn’t input—it’s a pretty powerful tool. Everyone worries that if we invent a super AI it will take over the world. But where would it get the idea to do that? It doesn’t have any drives. We as humans have drives, to procreate, to stay safe, to relate to people who are in similar tribes to us because those drives are very adaptive.

We have a tribal drive to find people who are like us. We have other tools, like AI, rockets and social media that we’ve invented that are also very powerful. I suppose we use them to fulfil our ‘monkey’ drives to connect to others. So, will a computer want to take over the world? No. Not in and of itself. It has no drives, that’s a very human thing to want to do. Though, we people, unless we’re careful, we’re going to make it worse for ourselves not better.

So the limits need to be on ourselves, and what we let ourselves do with our new tools. That’s as true for AI as it is for other things like social media and mapping the genome.

It’s not only tech companies that require tech talent. Having interviewed over 500 candidates during your career, what’s your advice to those companies wanting to find top tech talent?

Be a good place to work. I think that’s the most important thing. There’s a tech shortage, so if you have certain skills and on top that you can work in a team and collaborate well, you can go kind of anywhere.

You can post a job and you might not get any applicants, because who are you? You need to make a case. People with in-demand tech skills are just like any other humans. They want to work on things that are interesting, in teams that are fun, they want projects that are technically challenging, but also want to know they’re doing something meaningful, like ‘helping people get jobs’. That sells! Indeed’s mission is a mission people get.

It isn’t just ping pong tables and bean bags that will attract tech talent. You also need to treat people well during the recruitment process. Set the right expectations, interview well and keep communication flowing. If you don’t, you can lose people in the funnel. People with certain tech skills are in demand and aren’t going to wait around if you take too long to decide.

You have a degree in AI and Cognitive Science, but what did you want to be when you were a kid?

I always wanted to make stuff. As a kid I would make anything—board games, toys. When computers came along, (I was seven when I got my first computer) you could program them. So, I started typing in games from magazines that showed you how to program. I made my own little chatbots and things. I thought it was so cool that you could just talk to your computer, or have a little conversation with it.

I tried to make a translator, but it wasn’t any good. I also tried to make a cryptic crossword solver and that was much harder than I thought.

If you could apply AI to solve any problem, what would it be?

Education and healthcare. Teachers spend so much time marking, but all the important interpersonal stuff doesn’t get as much time. I wonder if there’s opportunities there to free up high quality teaching time. Also, in medicine to detect problems earlier.

Is it the responsibility of employers to re-skill workers if jobs are replaced by AI, or individuals?

I’m not sure. Singapore is doing some innovative things in this space though. It’s called ‘SkillsFuture Credit’. Singapore doesn’t have much of a social safety net (such as the dole). But what you do get is credits to be re-skilled. So there’s a whole training system where you can learn different things and it will help you figure out what things you’re interested in and what skills might be valuable. They’re working with different industries to help understand what skills they need. This is an experiment, there may be other programs out there I’m not aware of.

Are there any particular skills people should be focusing on for the future?

Working with other people seems to be pretty key. Being able to persuade and relate and mind-read a little bit (know what the other person really means when they say something) and work collaboratively. We keep being worried that everyone is going to be put out of work, but this isn’t a new problem. Every change does displace people, but so far we haven’t all been replaced by robots.

Use Indeed to find your next hire