Podcast: Meet the AI That Helps You Ace Job Interviews

In this episode, we discuss how 2 Carnegie Mellon University graduate students developed an AI system capable of giving you feedback on your interview performance in real time!

This podcast is sponsored by Mouser Electronics. 

EPISODE NOTES

(3:45) - Emotion detection system puts a smile on their face

This episode was brought to you by Mouser, our favorite place to get electronics parts for any project, whether it be a hobby at home or a prototype for work. Click HERE to learn more about how computer vision and AI are being leveraged today for defect detection use cases like preventing a bad apple from making its way to your local grocery store!

Become a founding reader of our newsletter: http://read.thenextbyte.com/

Transcript

What's going on friends? Let me ask you a question. Have you been frustrated with the way things are in the job market? Well, you're not alone. It's pretty tough out there and doing interviews takes a toll on you. But these folks over at the Carnegie Mellon University have come up with an AI solution that can actually give you feedback in real time. So, button up that shirt and let's meet the AI that's going to help you ace your job interview.

I'm Daniel, and I'm Farbod. And this is the NextByte Podcast. Every week, we explore interesting and impactful tech and engineering content from Wevolver.com and deliver it to you in bite sized episodes that are easy to understand, regardless of your background. 

Farbod: Alright folks, like you heard, today we're gonna be talking about how we can leverage artificial intelligence to analyze the emotions of human beings. But before we get into that, we're gonna talk about Mouser Electronics. Now, if you've been a fan of the podcast, you know that Dan and I, we're big fans of Mouser Electronics. Now Dan, why is that?

Daniel: We're fans of Mouser, not only because they're just an awesome electronics distributor and supplier, which they are. They are? Love to use them for all of our electronics projects at work in it and in our personal lives. But on top of all that, the way that they're connected to these awesome folks in industry that are working on cool things, developing new technologies means they understand the trends in new technologies, very similar to the way that we feel like we do on the podcast. Always talking about what's new, what's impactful, what's interesting. Mouser is the same thing has awesome technical resources, and we're all about sharing and spreading knowledge on the pod. So, we're all about sharing and spreading the knowledge that Mouser has shared and spread with us.

Farbod: Yeah, you know, as like hobbyists and tinkerers and engineers, obviously we love to learn how things work, what makes something tick, right? And what's been cool about Mouser is that they consistently put out fascinating resources like this article that we're linking to the show notes that talks about leading edge technology or something that you might take granted, like how your produce is getting started and all the engineering that goes on behind the scenes. So that's one of the reasons we love working with them. And just quickly about this article, it's an overview of how your produce, which means your fruits, your vegetables are getting sorted at a very, very rapid pace. And all the AI algorithms that are leveraged to make that work and the little robots that actually do the sorting. And basically, the miracle that goes from the produce reaching the factory to the plate that's at your table.

Daniel: And to make sure you don't end up with a rotten apple.

Farbod: Exactly. Or a green tomato. Right. So that was pretty exciting. And for me, at least one of the best parts of the, about reading these mouser resources is getting that little bit of inspiration of like, like, could I do that? Like, could I make a project like that myself over the weekend? And if, if you find yourself in the same place, well, Mouser is actually the one stop shop because not only they inspire you, but then you could buy all the components and find the schematics that you need to make that happen.

Daniel: Yeah. They, they give you the schematics, they give you the bill of materials. And oftentimes they even give you like the starter code to get going. I mean, it truly is your one stop shop for knowledge, for parts, et cetera, related to cutting edge technology.

Farbod: What more could you ask for as a tech enthusiast?

Daniel: Nothing.

Farbod: Yeah. So, folks, if you are feeling like you got a green thumb this weekend, or if you're just excited about AI and computer vision or just technology in general, check out the article. Now, that is a good segue to talk about the meat and potatoes of today's article. And we're going to Carnegie Mellon University, right?

Daniel: Yeah, and it's this team of, I think, master's students. Yeah. Cai and Agarwal, and what they're working on is a project for the IEEE Rising Stars Conference, which is...

Farbod: Wait, what does IEEE stand for? International Electrical Engineering? We should know this.

Daniel: I don't know man. Institute of Electrical and Electronics Engineers.

Farbod: Okay, you know what? I got what, 50% of the words right?

Daniel: Yeah, you missed electronics, but.

Farbod: Darn.

Daniel: Close enough. What I will say is we grade acronyms that we encounter on this podcast. It's a habit of ours.

Farbod: B for me. I triple E gets a B for me.

Daniel: B minus.

Farbod: Pretty, really.

Daniel: Electronics and electricals, pretty close. You could just be.

Farbod: But then you'd be I double E. I don't know how I feel about that. I feel like the triple E really drives it home.

Daniel: Yeah, you're right.

Farbod: But I guess what I was trying to say is that it's a professional organization for electronics and electrical engineers. And so, these students, they were trying to get a project ready for this conference. Now, Daniel and I, we haven't been to an IEEE conference, but we've been to an IMEC e-conference, which is essentially the same idea, but for mechanical engineers. And these things are massive. You have people coming in from all over the world to talk about the discipline that you're doing. And you have undergraduate research, masters research in all the different niches on the sub domains of whatever the topic is, in this case, electrical engineering. So obviously as a student, you want to pick a project that's really going to drive home that that's going to stick the landing for your audience. Right.

Daniel: And in this case, right, this rising stars conference is focused on. Like young professionals, students. And so, they're like, what to all these young professionals and students have in common? It's the fact that they're beginning their career, they're job seekers, they're young professionals, they maybe don't have a lot of connections. So, what's the most important part when you're a job seeker, applying for jobs, waiting to hear back. The most important and most critical part for these students, these young professionals at this Rising Stars Conference in their mind is interview and interview preparation. It's really challenging unless you've done a ton of interviews to feel fully prepared for a technical interview with someone. You want to leave a great impression and it's kind of frustrating. I think mostly for engineers who struggle a lot with communication to feel like they could succinctly communicate and also portray their emotions in a way that it leaves the other person excited to hire them and bring them on the team.

Farbod: Absolutely.

Daniel: And it's really challenging unless you're like super awkward and you ask your friend to sit there and ask you interview questions or you ask your girlfriend to sit there and ask you interview questions like I did, like it's really challenging to be able to prepare in a way that you get real constructive feedback saying like, oh wait, you actually make this weird face when you're thinking that makes it look like you're pissed off. That's not a good thing to do while you're on this video interview or in this in-person interview. You might give the other person the wrong impression. And so, what this team did is they tried to develop an AI driven system to help all these young professionals get better interview preparation, provide feedback on performance, essentially make the AI that helps people ace job interviews.

Farbod: And they really hit the nail on the head here because I know I would have truly appreciated a system like this. When I was a what, a sophomore or junior looking for internships, or even when I was a senior and I was applying to full-time jobs, because there was nothing of the sort. You're absolutely right. You either gotta find a friend or a girlfriend that you can sit down and do those mock interviews with, but there is no other resource that can kind of replicate that environment for you. And it's super anecdotal, but like, I remember my first interview and just how abysmal it was. It was so bad in fact that it's like a core memory ingrained in my mind of just how badly I did because I wasn't ready for it. And now you're offering students with a tool that will ease them into it and make sure that whenever they do the real interview, they're as comfortable and confident as they can be.

Daniel: Yeah, and honestly, the most interesting part about this for me, is they were able to achieve it with a technology stack that they didn't have to reinvent the wheel on this. The building blocks, to some extent, already exist, right? They took the software engineering class, it started as a project in their software engineering management class, who's led by Professor Catherine Fang, I think, who happens to be a professor, but also a successful entrepreneur. So, it's like, I appreciate the academia plus entrepreneurship aspect there that probably challenged these folks to make a really awesome and interesting tool that actually has real impact in the real world. But they had machine learning models and computer vision similar to the ones that we mentioned in the Mouser article at the beginning of this episode that already exist. They have these building blocks, these toolkits for them to build with. And one of the hurdles they encountered was getting the correct data set, like a large diverse data set of people with different faces, different colors, different backgrounds, to get this diverse data set, to be able to label and train images for their model successfully. They, I think there was a guest speaker in their class from this company called Hume AI, who was able to help them access this really extensive data set via an API. So, they basically had all the tool kits sitting there, but that's not to diminish the importance of their development here. They made this tool that provides real-time feedback, unlike traditional interview practice tools maybe like record a video of yourself interviewing and upload it to Reddit and be like, hey, roast my interview. Or to watch it yourself, like you can actively look at this. And I encourage people to check out the link in the show notes so they can see this, you can actively look at your face on the screen and get a bunch of indicators like, oh, you're appearing to be interested or happy or excited. And it also like, because of this diverse data set that they use, they feel really confident that across various ages, genders, and skin colors, the system can still very accurately assess emotions and give you real-time feedback on how the other person on the other end of that interview might be receiving your face. And I think that technology all on its own is really interesting and awesome that these master students were able to develop that. But even more so to be able to tailor that specifically towards job interviews, I think it shows that maybe you don't need to like reinvent new technology from the ground up. It makes me feel better. Cause you don't have to be like a mad scientist engineer to reinvent stuff from the ground up. You may just have to be like, have your eyes open about being able to apply the existing technology in a new way at the right time for the right people. And you can make a huge impact.

Farbod: Well, I think at our core, right. Engineers are creative problem solvers. Yeah. Right. And being able to identify a problem that just might be missed by others and knowing enough about your field to understand what it would take to address that problem and bring the solution together. That is engineering at its peak in my opinion. And if you don't have to create something else to go on top of it or invent something new, that's just a cherry on top because it means you can deliver much quicker than before. So, I'm going to take this idea, which I think is fantastic as is, and add another layer on my wish list, which is imagine like you go on LinkedIn, right. And there's a job posting that you're interested in applying for. Imagine that this tool could not only analyze your face, but could also use an LLM to parse through the job description, come up with questions, ask it of you, and in real time, test how you answer questions in a pressurized setting, like how do you do under pressure?

Daniel: No, I think that's a great idea. Right. You can connect this with a large language model to get tailored interview questions related to the job description. You can even have it grade your verbal responses as a response to those questions. How well are you proving from your experience that you measure up versus the requirements in the job description? And at the same time, are you doing it in a way that like portrays the proper emotions? Do you look angry and disengaged or do you sell yourself as someone who's trustworthy and happy and excited about the role. I think that that's super interesting and I appreciate you taking your, you know, this wish list and saying like, here's, let's plug this in and make this even better. I had to take a completely different approach. And I thought of like celebrities who get PR coaching. Like Nelly actually took some PR classes in college and she would always tell me about these crisis management courses she took. And they taught them how to respond in press conferences under tough questioning. And I think they even had them stand up in front of the class and the professor would ask them a tough question like, oh, did your client cheat on their wife? And you have to sit there and answer the question and satisfy the question asker in a way that also doesn't betray your client. I think that this could be really interesting. I feel like there's a lot of celebrities out there. And honestly, people in general who just respond really poorly when they're asked like man on the street style questions that catch people off guards. I feel like you could use this as like…

Farbod: That's a fantastic point. I'm picking up what you're putting down. That's a great idea.

Daniel: Create like a consumer application. It's like, this will just coach you to be a more jovial and likeable person in conversation period. Cause I truly feel like if you can sell someone in a job interview, you can probably sell someone in a blind date. You can probably sell someone on the street.

Farbod: On a car.

Daniel: Like it's, I feel like job interviews, you know, depending on like how selective the company is, maybe they're just clamoring. They're hoping anyone wants to work for them. But they in a job posters market, acing a job interview means like you're smooth talker. And I feel like the same technology could be applied if it's helping you ace job interviews to just help, I don't know, improve people's social skills. Period.

Farbod: I totally agree with that. And niche application of that, I guess that I just thought of because of the PR example you gave, I've been watching suits nonstop, lawyer training, getting them ready for defense or like putting someone in the hot seat. This would be a great tool for that.

Daniel: Not only that, but also like, if you could somehow live monitor the jury.

Farbod: Yeah, and see how, oh my God, you're right.

Daniel: How many of the jurors do you feel like are like in your pocket?

Farbod: Or resonating with what you're saying. Damn, that's a great idea.

Daniel: Man, they've got a lot to work on.

Farbod: Yeah, I know, we're just just keeping adding to this wish list, we could go on all day.

Daniel: I'm sorry, Cai and Agarwal, we were so excited about your technology. We've created a bunch of new things for you to develop, but I think that's just it goes to say how impactful and the awesome potential that this technology has, and then also the way that. This it's awesome. Just this team of students was able to strike while the iron's hot, develop this awesome technology at a time where the technology is mature enough to be able to actually make a difference and in a domain that can impact so many people. I think one of the things we promised we would do is purposefully mention like the pros and cons of each technology from an impact perspective. And I just wanted to note those before we wrap up today. One of them being the pros, real tech that actually works to provide personalized real-time feedback based off of your face, uses comprehensive data set, so it's able to be accurate across a diverse set of subjects. And it can be beneficial for various applications from job hunting to speed dating to PR responses to even in the courtroom. Cons, I would say it, it may require significant computational resources to run. I have no idea how heavy this is to run the model. Similarly, I think they have yet to truly test the limits of the quality and diversity of the data set. Maybe for some people it's really, really good at telling whether they're happy or sad or being perceived as happy or sad. And maybe if other people have beards on their face versus people who've got you know baby faces like me, if you've got a beard on your face maybe it's got trouble detecting your emotions and maybe I'd feel like that's just something that comes with more testing and more trials.

Farbod: Or if you're very you know animated as a Persian man that might show you off as too excited for a situation where you're not supposed to be excited and maybe not taking that and take out.

Daniel: Or if you're cool calm and collected maybe they say you're not engaged.

Farbod: Exactly! Exactly! But just to wrap it up, let's do a quick TLDR. So, we got these two students from Carnegie Mellon University who were getting ready for this conference at IEEE, which is for electronics and electrical engineers. They had to come up with an idea that would really resonate with the core of that crowd, which is, you know, up and coming engineers think, other students, fellow students, and they were like, what is a problem that they all struggle with? Well, for most of us, it's finding a job. And doing an interview is a core part of that process. So, naturally with the boom of AI, they were like, how can we leverage this technology to make the interviewing process a little bit easier on these folks? How about we could take the AI and have it analyze their face and the emotion that is being perceived by the interviewer while you do a mock interview? That is the core of what they developed with this awesome company called Hume AI, which helped them at every step of the way using their APIs so that they had a lot of training data for people of different skin tones, genders, etc, to make sure that this model they came up with was as flexible and diverse as possible.

Daniel: Nailed it!

Farbod: I do what I can, but Cai and Agarwal…

Daniel: Did way more than we can.

Farbod: Did way more than we can. And with that, I think we're good to wrap it up.

Daniel: Yeah, let's wrap it up.

Farbod: All right, folks, thank you so much for listening. And as always, we'll catch you in the next one.

Daniel: Peace.

As always, you can find these and other interesting & impactful engineering articles on Wevolver.com.

To learn more about this show, please visit our shows page. By following the page, you will get automatic updates by email when a new show is published. Be sure to give us a follow and review on Apple podcasts, Spotify, and most of your favorite podcast platforms!

--

The Next Byte: We're two engineers on a mission to simplify complex science & technology, making it easy to understand. In each episode of our show, we dive into world-changing tech (such as AI, robotics, 3D printing, IoT, & much more), all while keeping it entertaining & engaging along the way.