DUE TO SOME HEADACHES IN THE PAST, PLEASE NOTE LEGAL CONDITIONS:

David Yakobovitch owns the copyright in and to all content in and transcripts of The HumAIn Podcast, with all rights reserved, as well as his right of publicity.

WHAT YOU’RE WELCOME TO DO: You are welcome to share the below transcript (up to 500 words but not more) in media articles (e.g., The New York Times, LA Times, The Guardian), on your personal website, in a non-commercial article or blog post (e.g., Medium), and/or on a personal social media account for non-commercial purposes, provided that you include attribution to “The HumAIn Podcast” and link back to the humainpodcast.com URL. For the sake of clarity, media outlets with advertising models are permitted to use excerpts from the transcript per the above.

WHAT IS NOT ALLOWED: No one is authorized to copy any portion of the podcast content or use David Yakobovitch’s name, image or likeness for any commercial purpose or use, including without limitation inclusion in any books, e-books, book summaries or synopses, or on a commercial website or social media site (e.g., Facebook, Twitter, Instagram, etc.) that offers or promotes your or another’s products or services. For the sake of clarity, media outlets are permitted to use photos of David Yakobovitch from the media room on humainpodcast.com or (obviously) license photos of David Yakobovitch from Getty Images, etc.

Welcome to our newest season of HumAIn podcast in 2021. HumAin is your first look at the startups and industry titans that are leading and disrupting ML and AI, data science, developer tools and technical education. I am your host David Yakobovitch, and this is HumAIn. If you liked this episode, remember to subscribe and leave a review. Now onto our show.

David Yakobovitch

Welcome back listeners to this episode of HumAIn. Today, we’re talking about not only the future of work, but the future of diversity and how to leverage AI to build smarter, better and stronger teams with humans first. Today on our show, we’re bringing you the CEO and founder of eightfold.ai, Ashu Garg. Ashu, thanks so much for joining us on the show. 

Ashu Garg

Pleasure, David, super excited to be part of HumAIn podcast. Thank you for the invitation.

David Yakobovitch

I’ve been following Eightfold’s story for quite a while, and it’s really exciting to see the growth that you’ve experienced as a venture and continuing to build around talent and diversity. Before we go to where you’re at today, can you set the stage for our audience about yourself, eightfold.ai, and what led to the creation of your venture? 

Ashu Garg

Let me start by sharing a little bit about myself. What got me here, I grew up in the foothills of Maliya in India, but my undergrad from IT in India came to UIC to do my masters and PhD in machine learning and AI. Back in the days I was at IBM research followed by Google research, started to complete, we’ll meet after that. And in 2016, we started at eightfold. It’s something that, well, a participant in the study, setting about that one, but as I reflected back on my own story, it was really a series of good plus, helping hands from people I knew that is what got me to where I am. 

Then around 2015, 2016, as I started thinking about what’s next for me, a few things happened. I realized how important employment is for people. If you have the right job, it can change not only your life, but everyone around you. Because while I got lucky, I had access most people in the work don’t, and through data, through here, actually, I can change that. 

Because with machines at the scale, we can understand who can do what. And third, both as an entrepreneur, an early employee at Google, realized how important talent is in the success of any company. If anyone comes and tells you otherwise about the team at the company, they’re just kidding themselves. I am where I am because of the team that I have.

So as those things came together for me, the importance of employment, our data problem and importance of talent enterprises. It became obvious that eightfold is my calling. We have a talent intelligence platform that is being used by the leading enterprises across the globe to hire, engage, and retain a diverse workforce.

David Yakobovitch

When you look at tools and technology, they’re always changing and always growing, but it starts with people. It’s, as you just described, Asha, it’s the people on teams that make us who we are and scale our organizations. A classic phrase, one plus one is greater than the sum of the parts. In fact, it’s three, and it can be many times that multiple as teams scale and they are synergistically working together. When you think about the world today, and as we’re moving into this hybrid world of in-person digital experiences, what do you think about the ethics, the responsibility for business operations, for building teams that are people first? 

Ashu Garg

We are cautious, there is research that shows that there are teams that perform better because they bring a different perspective, as cruel as it gets. But more fundamentally actually today, there is no other alternative. If you go and talk to large enterprises, everyone will tell you their number one challenge is people. They are not able to hire fast enough. Now in that world, if we keep saying no to 70% of the population: women, people of color, people who are in the second half of their life, like getting closer to 45, 50, 55 year old. How are we ever going to solve the talent problem in our society? So to me, not only is it important for businesses to succeed, there is no other alternative, they have to do that. So it has become extremely important for enterprises to think about diversity, to think about their own biases, to understand what talent exists. We added exits to bring the right people on board and that is where data and AI comes into play.

David Yakobovitch

Now when we talk about the talent problem, I’ve often heard of it in education as a pipeline problem, but is it a pipeline problem or is there something more to it.

Ashu Garg

My plan is the most obvious, simplest, easy answer. I don’t have enough people in the pipeline. But more importantly, quite a few times, we are not looking at the talent in the right way, whether it is about understanding people’s backgrounds or it is about understanding their potential. So, for example, you are in the Bay Area. Everyone is talking about hiring people from Stanford and Berkeley and MIT. That is all goodness. 

What about Fudan University in China; Bit’s planning in India; Technion in Israel? and so on. These are some of the best qualities in the world, but we’re not even familiar with them. So we ended up saying no to those people. Second is what has happened over the last 20, 25 years, technology has seen exponential growth. What I tell everyone is that iPhone on Android smartphone that you have in your pocket, that practically didn’t exist 10 years back. That is the concept we know today is going to be activated in the next four or five years. So the average half life of a skill is less than five years.

The rate at which they come into the market and disappear is faster than ever. Are you interested in a joke? But actually the scenario that happened with us when we started the company, one of the managers in 2016 asked me, from another large enterprise: Ashu, Can you help me find people who have five years of experience in Golang programming language?

Sure. Let’s do it for five years and I’ll help you find that. But the essential is this. We can’t keep looking for people who have done the work. We have to look at the people who can do the work, and that is a fundamental shift in the mindset. But once you change that mindset, you can look at the people from a very different lens. So it’s no longer a pipeline level. 

That is how you will solve diversity. Because as an example, in the Bay Area in tech, less than 20% of the engineers are women. So if I just keep looking for women who are in data scientists, I will never solve the problem. But on the other hand, if I started looking for a woman who can do that one and then lead that goes from 20% to 40% and have productivity in no time. That is what eightfold is all about.

David Yakobovitch

I was having a conversation, a few weeks ago, with one of the leading psychologists working with DEI and working with large multinational institutions. The work that she does, she told me that some of these executives would come up to her and say: Am I a bad person because I haven’t been thinking about DEI for so many years, because I haven’t been intentionally trying to hire people from different backgrounds, and I haven’t thought outside my narrow world view, am I a bad person for that? What would you say to executives like is now the time to change or  now’s the time for a new opportunity? 

Ashu Garg

I think we need to be aware about our actions and the implications as an example, ignorance can’t be the answer. So, I don’t know whether you’re a bad person or not, but… We have our conscious biases and we have our subconscious biases. I believe that most people these days don’t have much conscious bias. They are trying to work around that, but subconsciously we are, and we need to proactively work on that stuff.

We need to reach out to the people who may not have had all the privileges that we have and support them. We have to look at people beyond what we perceive for  their face color, daily age. Unfortunately in the Bay Area, if you have 50 plus it’s very hard to find a job as a software engineer. We have to start looking past that. And it all starts by understanding that we have limitations and many of your audiences, David, people who understand data, AI, computer science. I will give you one interesting perspective among multiple big differences between machines and humans.

One of them is that machines have the ability to forget and ignore. But with humans, it’s very hard for us to forget things. They just get edged and we end up paying attention to the things that don’t matter and once we realize that  is the case. Second is that we have our biases coming because of the lack of knowledge, where that example I was giving you about Fudan University in China, all about a certain skill mapping. If you’re looking for a machine learning person and you’re not able to find one, maybe people with economics background or signal processing background could be a great candidate for you.  Which is a data platform. Knowledge. So actually knowledge and  moving out of biases can really help us solve this problem. So maybe you have not done it, but it is just the time to fix this problem. 

David Yakobovitch

We’ve seen quite a  few solutions in the industry to attempt to solve for bias. I recall a few years ago there was a big tech company that had launched an HR tech solution that was trying to find the best candidates, and there was a flaw with this system that came to the market. It seems that women were excluded from the interviewing and hiring process because it was completely run by a machine, completely run by the AI without any supervision or oversight by humans. Sounds like the technology was just planned to run its course, but perhaps you have a thesis that there needs to be a human AI augmented system.

Ashu Garg

Part of that is the same ignorance that I mentioned. It’s not being blind about the problem that you’re solving. You have to be proactive about that. Are you using the data set that it is to begin with very biased? Audio feeds are being extracted and are catching those biases? or your algorithms are not optimized for the challenges that an imbalanced dataset will help.

Because again, if you’re trying to train your avoidance on data scientists on day one, what you will see is that the women are underrepresented over there, but you need to adjust for that so far. Finally, there has to be an audit process to ensure that your algorithms are not going crazy and that are doing the right thing. Then at the end of the day, it’s not magic. Let’s use it to help humans do a better job. 

David Yakobovitch

I’ve often read that Carnegie Mellon has their ethical design framework from their software engineering school. And what they have here is about human machine design teams together. They have this checklist where it’s making sure that you have human oversight? Are you having an audit? Are you having observable features? Can we explain it with simple language? And these are some of the contexts to think about.

I know at our startup today at a single store, we’re talking, also, how to reduce bias in hiring, things such as removing the candidates name, such as, even anonymizing a voice. There’s a lot of technology that can assist in the process, but it sounds like it’s more than the technology, at the root, It’s a human problem. 

Ashu Garg

It’s all about humans. What makes it hard is hiding it with some mechanism. Ultimately, you’re trying to bring two people together, who are, eventually, going to work together and you need to give each other that comfort as well. But this is what I’ve set up to hiring method of a good candidate for you, why this is a good commute for you around transparency.

Even though this person may not look and feel like you. They have all the things to start to make them successful in your organization. And did that gain yet that, yes, this is a good place, and this is the place where you will succeed. And really thinking from the candidate perspective and that honeymooners perspective. I think we at the algorithms and systems have failed when they have succeeded many times. These systems are designed to just come in and replace humans. In that case, not only you’re taking the snitch system correctly, you’re teasing that: I really don’t need to worry about humans, and that has to be front and center. What I liked, David, about your podcast, It’s all about  being human. So, how do you bring the human thing? 

David Yakobovitch

And as we look at the data points, you’re completely right. Ashu. Because you can look at candidates from Carnegie Mellon and Stanford and MIT, and say that you want the creme of the creme. But frankly, if you look at someone with a 3.9 average master student from Stanford and 3.9 master student from University of Florida, they could have completely the same skills. In fact, maybe even the engineer from University of Florida might be the better Golang programmer, but you wouldn’t know if you gave additional weight to the university ranking.

Ashu Garg

Absolutely, and there’s another big park over here. Quite a few times we treat this as a uni-dimensional problem. We say that: Oh, who is the good person? Who is this great engineer from Stanford? The challenge with that is you are not taking the context of the work into account. Not everyone is the right fit for every person in every job.

One of the things that we believe very strongly at eightfold is that it’s not that people are good or bad, or one is better or worse, who is the best fit for which flown in that company. I can’t have that engineer from Stanford with 3.9…, because they may not be a good sales person. They may not be a great product manager because humans are very complex.

Each one of us has thousands of skills in that role that you’re hiring for people today, whatever role you’re hiring for that role, really what over the next three, four years. So once we started thinking about the fit, what system best, everything changed. 

David Yakobovitch

Today, I do a lot of startup advisory with early stage ventures and there was one startup that I was advising. They came out with some new job requisitions for intermediate and senior full stack engineers. And they said: Hey David, can you review the job? So I said: Sure. And I read through it. And I said to them: I think there’s some potential issues with your job rack that are going to exclude candidates.

There were things like: must have a Macbook computer, or if you self rate yourself on this skill, you rate yourself 9 out of 10 or some other things like that. And I said: Well, these are things that could exclude anyone who develops on a Windows machine or someone who, a lot of women in the workforce, as the research would show, may not always rate themselves 9 out of 10, but may actually be better than some of the men who rate themselves 9 out of 10. So it’s important in how we generate the context for the search to be fair so that everyone feels welcome to apply to jobs as well. 

Ashu Garg

Absolutely. So one interesting thing, back in the days, I did some research in the area of opinion pooling and the basic thesis over there was, when you say 9  out of 10 and when I say 9 out of 10, do we mean the same thing? Some people like to read  everything in midair, some on the extreme. So even this concept of rating that’s not good to get too far. So you have to really assess the people at their full potential. Another thing I would say that happens in many of these setups, which makes it very hard and tricky, is we give too much weight to one hour interview process, taken in the early days of starting the company. Someone told me as a hiring manager that: Ashu, I don’t care what people have done, what their GPA was, what schools they went to, what they have done in the past. I will do the evaluation by myself in one hour and see if they are good or not, and that also leads to a lot of biases. 

We are saying that we won’t care about all the work that you have done for the last 10 years. How well you do? How well you execute? How will  you get it? What we’ll do is we will take the next one hour to figure out everything ourselves. Frankly, I’m not smart enough to evaluate anyone in an hour, and to be honest, I don’t think anyone is.  But once we started looking at people’s execution in the past, we also able to get out the biases very quickly. For any meeting of one hour I will have a lot of biases.

David Yakobovitch

At my heyday of mathematics, I was ranked number 15 in the United States, in the calculus division where I did these competitions. I was this young whippersnappers. I could do all these problems super quick and I was impressed and people were impressed. But today I wouldn’t be able to do that.That was me years ago, today I use programming and code to do that. The same analogy can be true for these one hour code challenges.

If you put someone up and say:  Okay, let’s see if you can build this in C ++ or Go or Java in one hour. Well, sure. If I spend a few hundred hours coding to do it, I could probably do it. But is that testing for the people you really want to hire for the job?

Ashu Garg

Maybe we should talk to you about the job, but that is exactly the case. That is exactly the case, you’re testing for the wrong skill. In fact, I remember during my PhD, in one of the exams, one of the students asked the professor: Can we bring a book to the examination hall so that we can just refer to it as we are doing the exam. And his answer was quite interesting. No, you just need to solve the problem. If you can solve it that’s good enough.

But during the activity today, what we do is we completely change that. You’re like: Oh, you can’t use Google. You can’t go to Stack Overflow or any of these places to learn about it. You have to just answer on the spot, quote on whiteboard without a compiler, without an editor. That’s not the real world, then you wonder why this person is not performing when they join, because you were testing for the wrong things.

David Yakobovitch

In your opinion and experience, Ashu. I’ve been involved with several startups and a lot of students who I’ve mentored as they’ve gotten careers in data science and software engineering. I hear from a lot of these students: It’s not fair. The system’s broken. How do they get this way? What do you think about the state of the system today for these engineering interviews?

I know some companies are doing it right. They’ve actually completely scrapped those one hour code challenges, but not everyone has. What can we do to be more equitable with our engineers?

Ashu Garg

I did pull that is what our attempt is, I See these interviews at some level, in fact, even worse. I’m sure you have heard  that sometimes the interviewer It’s just trying to prove that they’re smarter than the candidate and satisfy his own ego. But under the hand, that’s not offensive to anyone. What happens today in the way the systems are designed. As a hiring manager I get one patient as you and me. I have to read that resume  and try to understand this candidate.

If this person went to Stanford and I also went to Stanford, I understand that. If this person was in the search quality team at Google, as I was, I can understand your work, But on the other hand, this person worked at this company that was running something with Docker containers that I’m not too familiar with, I don’t know what to do. Then All I can do is put them through an hour, two hours of grueling interview to validate them. Is also a real challenge for people to be fair to them. What we are trying to do through machines is help hiring managers understand that candidates past, be able to dig deeper with you, look at the peer group of the community to see what their peer group is doing today.

Look at the career trajectory of that candidate, to help the hiring manager understand how this candidate has fared compared to others in his or her prior company. That has dramatically helped us reduce some of these biases and provide a better experience. 

David Yakobovitch

Recently I was working with another venture where I was helping them with their executive search for their CTO. Initially the company was only looking for candidates in their twenties and thirties, because they wanted to be really startup focused,  and I said: You should broaden your search, look for people in their forties and fifties. You may be able to find someone very talented who’s been there building systems before. When they opened up to that level, they found them an incredible candidate, who’s very humble, who’d been an engineer for 25 years. Built many systems, and now it’s been a great, fantastic fit. I think about small success stories like this, also I think about the sad stories that we’ve seen in the news the last few years, where other large multinational companies decided to provide early severance for employees over the age of forty. For I’m not sure the reason, I can say, because there’s so much business expertise and software expertise that people build up throughout their careers that I think you’re right. It’s often not measured in the interview process.

Ashu Garg

The thing is that we miss that many of the jobs are not only about coding, but about people to having that perspective on the problem. I don’t have all the details, but I still remember  that when I joined Google, at that time there was this individual who was the first repeat engineer at Google, but he joined Google post retirement from his current job.

So he was not the 20 year old typical person. He was already working for 30 years plus and retired,  and then he joined. So some of these success stories of the companies that we know out today in the world, the reason they’re being successful is combining that experience with the young talent. If companies pay attention, focus on that, how do we bring everyone together? And one thing that I’ve done in our company, which I’m super excited about, is that many people with lots of experience combine them with the people who are just coming out of college, and what it does is that brings two way different perspectives together. It challenges both sides really well. The fresh thinking coming from these young people, combining that with the experience of the other leadership team and to meet that combo sometimes works really well.

David Yakobovitch

As we’re planning the world post pandemic, we’re going back into a hybrid world of not only remote, but also in the office. There’s been a lot of people who’ve been impacted both in tech and other industries, whether fresh college grads or seasoned engineers who are now looking for new opportunities. Whether some of the trends you’re seeing for the workforce as we go back to hiring, and also whether some of the feedback or advice you would share with candidates on the market today?

Ashu Garg

The next one year is going to be very interesting. For the last 18 months most people were completely hunkered down because of COVID. Not doing anything. And now all of these people are realizing that as the world is coming back to normal, what’s next for them. It is also giving them a perspective around what their company, their organization did for the people during these times.

So on the dead hand, there are many people who were impacted by COVID and they are now looking for a job. And then there’s a third group where people have now moved from one geographic location to another geographic location, and they may start looking for a job in another location. So I would say that the talent market rate landscape is completely going to go through a massive shift in next 18 months. As an organization, you should really pay attention to the people that you have onboard today.

Make sure that you’re taking good care of them. Make sure that you’re preparing for the chance that you may end up experiencing.  This is also a good time to hire great talent, because many people are looking up.  On the other side, if you are exploring, on one hand it’s a great time, but there is going to be a lot of demand and supply that is moving to the market rate. So take your time, think through what you really want, don’t rush into it. Because whatever you do, will define the thing for the next 5 or 10 years of your career. 

David Yakobovitch

With that feedback, you’ve shared a lot about what’s next for enterprises and what’s next for consumers in their search. If I may conclude with what’s next for eightfold.ai.

Ashu Garg

Well, we are just starting. We started the company with the mission of enabling direct credit for everyone in the world. I think there’s going to be a very long journey for us … So every day, our goal is to make sure people are cutting their employment. We are helping improve diversity at the workplace. We are helping everyone to get a job

David Yakobovitch

I’m excited to see as we continue moving back into the hybrid world and in-person world to get stronger diversity in the workforce., it’s been incredible to hear about, not only your story, but also your experience. Ashu Garg, CEO and founder of eightfold.ai, on how to leverage AI, to recognize and improve diversity in hiring. Thanks so much for joining us on HumAIn 

Ashu Garg

Thanks , David, I really appreciate this opportunity.

David Yakobovitch

Thank you for listening to this episode of the HumAIn podcast. Did the episode measure up to your thoughts and ML and AI, data science, developer tools and technical education? Share your thoughts with me at humainpodcast.com/contact. Remember to share this episode with a friend, subscribe and leave a review, and listen for more episodes of HumAIn.