INSEAD Professor Of Digital Disruption: ChatGPT Is ‘Worth The Hype’

Folks are still talking about ChaptGPT, the conversational-style chatbot which famously passed a Wharton exam. Since those original headlines in January, more has been written about the bot that provides answers to everything from difficult code questions to relationship dilemmas and where you can find Austin’s best dive bars — and even, in an infamous case, in the form of an Elon Musk tweet.

The impressive generative pre-trained transformer (GPT) is trained using large amounts of text data from the research papers, articles, literature and other information off the internet, including its apparent index of both Wikipedia and Twitter. Released in November 2022 by San Francisco company OpenAI, the chatbot’s launch triggered serious debate about its use for education and even its potential restriction. ChatGPT made headlines again when access was blocked from devices at New York public schools, which was seen as an unpopular move from some in the academic world.

Pretty much everywhere you look you read something about ChatGPT. Even existential questions transpired such as, if the bot will revolutionize the way people use search engines. Google just announced more details on the release of Bard – its AI language model rival to ChatGPT. Business schools may still be mulling over how or how not to deploy the technology in classes. And their final answers may vary. One Wharton professor told Poets&Quants in January its use was prohibited in his operations management class because the goal is to certify students’ skills. While another professor at the B-school, Ethan Mollick, is requiring students to use it in his classes on innovation and entrepreneurship.

INSEAD Business School Deputy Dean Peter Zemsky says there’s plenty of questions to be asked about ChatGPT, but it represents just one in a series of AI progressions over the last decade. First came image recognition and then AI’s ability to generate images and now, ChatGPT, the ability to generate text – an “obviously significant breakthrough new application,” he says.

Keys to understanding technology advancement are learning how or how not to use the specific technology at play. With ChatGPT, Zemsky says, the conversation – though a fair question – should be less about whether it’s used for cheating on homework or exams and more about it as a tool for improving business and society.

DAY-TO-DAY IN THE DIGITAL INDUSTRIAL REVOLUTION 

We are living through a kind of digital industrial revolution, Zemsky puts. Wave after wave of new technology has emerged and the progression of artificial intelligence (AI), he says, has exploded. A factor of this is the cost for processing, storing and sharing data has become relatively inexpensive.

“Fundamentally, what’s driving it is terabytes and terabytes of data and better processors and now, an ecosystem of people working on the algorithms to bring the processing power to data,” Zemsky says.

More businesses are hiring data scientists, trying to join this fundamental trend Zemsky describes and score-up massive profits that some Big Tech companies like Amazon and Google have generated. But Zemsky notes, despite burgeoning advancements, your typical company hasn’t gained a whole lot of value out the technology. In a digital-forward age, how to make a profit is what everyone is trying to figure out.

He teaches a popular online course at INSEAD on strategies for digital disruption, having also participated in the launch of the B-school’s online courses about a decade ago. His main ambition is getting students curious about the technology, hoping they might one day leverage it to drive value and efficiency in their businesses.

“We keep having these bursts of possibilities, some of them will turn out more than the hype and I think AI is one of those,” he says. “Others probably less than the hype, and that would be things like crypto.”

WHAT IS CHATGPT GOOD FOR?

Zemsky says ChatGPT or technology like it could be implemented in INSEAD’s immersive case studies. The immersion factor of the coursework is executed through live role playing the case, so rather than just reading and studying it, students may receive emails from characters to simulate real-life business interaction. Instead of receiving a standard email, ChatGPT could be trained to generate a personalized response.

“ChatGPT could respond to emails as if were Alicia, head of operations at the plant, and to actually play characters and bring that richness, so we are actively looking at this,” he says.

This idea is similar to AI deployed in Microsoft’s Viva Sales where sellers can use it for generating email content to send to consumers. Whether it’s emailing promotions, receipts or product information based on data specific to a client, the point, according to Microsoft, is to save the seller time while still communicating efficiently to customers.

THE ‘MAN PLUS MACHINE’ METHOD

Another example Zemsky draws up is the use of ChatGPT for summarizing information. Consider a student pastes his or her research paper into the box and asks the bot to generate a summary of that text. Warwick Business School is using AI technology for a fairly similar purpose. The B-school late last year released a tool to improve students’ writing called the AI Essay-Analyst. It generates feedback about students’ essays ahead of their final deadlines on issues of readability, sentence structure and referencing. In a survey, Warwick MBAs responded their struggle with writing posed the greatest barrier to their academic success, according to a news release from the B-school. The tool also reports how well a student links and explains ideas.

Isabel Fischer, an information systems professor at Warwick, says the tool is especially useful for dissertations that sometimes reach 10,000 words and could often be written during late nights.

“It is refreshing and invigorating to get a report with clear visualizations ‘on demand’, that reviews key features and gives suggestions for improvement,” Fischer says in the release.

The tool demonstrates both human and AI involvement, and Zemsky says this “man plus machine” bit is what companies like Microsoft have emphasized. Microsoft in 2019 launched a series of online courses called the AI Business School for educators to learn how to implement AI for business purposes. Zemsky partnered with the software maker to teach a course on strategies for driving business value.

HOW CHATGPT SHOULDN’T BE USED

Man plus machine is necessary even in the midst of vast innovation, simply because AI still makes mistakes. If certain applications are not checked by a human, they pose detrimental risks, Zemsky says. For example during Bard’s demo, there was apparently a mistake made. Experts in the field say machines can learn bias from its human creators and sometimes it can even overlearn and actually become more biased. Zemsky says AI algorithms like ChatGPT are “insanely complicated”, with billions of parameters, which means basically no one can check the algorithm. Sometimes AI might give bad advice. Sensitive decisions should always loop in humans. Don’t use ChatGPT for medical advice or questions like should I get the vaccine?

“Again, these algorithms learn from responding to the internet and everything that is inserted into that and not everything out there is correct,” he says.

That’s why he says if ChatGPT writes the summary of a research paper, its important a human validates it.

DON’T MISS CHATGPT JUST PASSED THIS WHARTON PROFESSOR’S FINAL EXAM. HE SAYS HE WON’T BAN THE AI TOOL and GO VIRAL, GET AN ‘A’: CONTENT MARKETING COURSE PUTS GEN Z’s TIKTOK SKILLS TO THE TEST.