You may not notice it, but AI permeates almost every facet of your life. From ride shares to chatbots to your carefully tailored Facebook newsfeed, machine intelligence has shaped how we interact with the world and with each other. In this rapidly changing landscape, Wharton Prof. Kartik Hosanagar plans to launch a new Wharton Online course to help students understand how AI fits into the business world. He called into the Wharton Business Daily studio to share his insights with Al Gardner, who was filling in as host.
Interview Highlights
1. AI has come a long way.
“One of the big breakthroughs was (when) IBM’s Deep Blue beat Gary Kasparov in chess in May of 1997, and that was considered to be a huge milestone. But since then, we’ve seen AI beat two-man champions in Jeopardy a couple of years back in Go, which is a strategy game like chess. Now, we’re no longer in this world of games. We’re talking about driverless cars and Alexa, Siri, and all these kinds of things. So yes, it’s no longer fun and games. It’s now real world and real life.”
2. AI learns best by consuming Big Data…
“In recent years, what computer scientists have realized is that a better way to build AI is by looking at tons of data and looking at the patterns in the data. What we might do instead of asking doctors, ‘Give us the rules to diagnose diseases,’ is to take data from over a million patients who have come into hospitals. We can look at the data on what symptoms the patient presented to the doctor and what the final diagnosis by the doctor was. We have statistical programs that are looking for patterns in there and are able to say, ‘Look, if it turns out that this person has chills and fever, and the fever has been going for over a week and a bunch of other symptoms,’ then we conclude that maybe it’s not a viral thing like flu. Perhaps it’s a bacterial thing and we give them an antibiotic. It’s just looking at the patterns in the data.”
3. …but this can be a double-edged sword.
“Reuters reported (that Amazon’s hiring algorithm has) a gender prejudice. There’s algorithms used in courtrooms to guide judges in sentencing and parole decisions, and that AI was reported to have race prejudice. If you look at why or where is that prejudice coming from, (you realize) these AI systems are trained on past data.”
“If the data is of poor quality, then what it learns is problematic, right? It’s garbage in, garbage out. It may be that data is not of poor quality, but there are implicit biases, unconscious biases represented in the data, because the data is ultimately how human decision-makers have made decisions in the past.”
4. AI can be creative too.
“We’re starting to see AI being used in those settings as well. You can see AI create short stories, AI write poems, AI come up with paintings, and you show it to human judges or human experts. You don’t tell them this is a computer. You just show them these creative outputs from AI, and it tends to get very high scores. We’re still in early stages of AI and creativity, but I’m very fascinated that AI is getting to do those kinds of things.”
5. How does AI fit into the business world? Learn more through a new Wharton Online course, “Artificial Intelligence for Business.”
“Just like digital technologies like the Internet have fundamentally transformed business over the last 20 years, I think AI will do the same over the next 20 years, and managers cannot afford to have a poor understanding of something that’s so fundamental to business today. This course is trying to address that.”
— Angela Lin
Posted: March 11, 2020
