Sramana Mitra: I still go back to the gender question though. It’s not entirely invalid. If you’re a man, unless you’re a transgender, you’re most likely not wearing women’s clothes. You may have experience with baby products, but as a man, you would not have experience with women’s clothing.
Ashutosh Garg: Do you want to base it on people’s gender or on experience?
Sramana Mitra: But how do you pick up the experience of somebody who has used a product?
Ashutosh Garg: That is where the data analysis comes to understand what you are doing.
Sramana Mitra: Here’s another example. I travel a lot. I love eclectic travel. If you look at my LinkedIn profile, I don’t think you can find that. However, if you’re doing employment search for a travel company that is looking for that eclectic travel experience, how would you find me for that job?
Ashutosh Garg: That is a very fascinating example. A few things. LinkedIn is just one of the many data sources out there. You have a lot more data on your blog than LinkedIn.
Sramana Mitra: Which brings me to another question. Does your algorithm draw from Facebook? I publish my travelogue on Facebook. Would your algorithm draw from Facebook or is that not part of the data source?
Ashutosh Garg: We draw from 6,000 plus public data sources. Your Twitter feed may be relevant if you’re being interviewed for a marketing role. If you’re being interviewed for a software engineering role, most likely your political opinions don’t matter. Maybe I can identify that you have been interviewing people across the world. You have been talking about multiple locations. That is what we are trying to infer.
Sramana Mitra: If I’m hiring software engineers for a travel company, I want people who are widely traveled or who are passionate about travel so that I can draw on that level of passion.
Ashutosh Garg: I think what you are saying is a common misnomer that we will have as a human being.
Sramana Mitra: Yes, I’m asking you how your AI picks up that information.
Ashutosh Garg: Our AI will see what are the characteristics of the people. Maybe it’s relevant; maybe it’s not. A few years back, what I was taught was to be a CEO, you need to be very extroverted. Research shows that most successful CEOs are actually not extroverts.
Sramana Mitra: Good CEOs need to be analytical and strategic. Those are more important characteristics.
Ashutosh Garg: In your example of travel. You can look at travel experience. Maybe it matters.
Sramana Mitra: On that one though, I would argue that it does matter. It’s a lot more fun to work on a problem or topic that you are passionate about. From a longevity of a hire, those things matter.
Ashutosh Garg: And if that is the case, most likely it will show up in the data, and the system will pick it up. I’m not trying to pass my judgement. Let the system learn from data.
Sramana Mitra: To net this out, we are still doing pretty much large-scale data and statistics modeling and not introducing truths from other sources into the algorithm.
Ashutosh Garg: Our customers may set a rule saying, “I don’t want people to switch jobs more than once every 18 months.”
Sramana Mitra: You can set those rules and the algorithms will take those rules into account. If you were not doing either of the two companies that you’ve been involved in lately, what problem would you be solving today? This is to provide pointers to our new entrepreneurs who are looking for problems to solve with AI.
Ashutosh Garg: The world is going through a massive transformation. We have a lot more data. We should think about how we can make the life of people simpler, healthier, and happier.
Look at the technologies and solutions that can fundamentally change people’s lives and make our society a much better place to live. Don’t just go after money but think longer-term whether it is around education, energy, healthcare, or employment.
Sramana Mitra: It was a pleasure talking with you. Thank you for your time.
This segment is part 6 in the series : Thought Leaders in Artificial Intelligence: Ashutosh Garg, CEO of Eightfold.ai
1 2 3 4 5 6