A friendly, distinguished-looking man appears on the Zoom call for the April Baker Retailing Board meeting. He intones, “Hello everyone. My name is Professor Ethan Mollick… I have been studying startups and entrepreneurship for over a decade and have some thoughts on the subject that I would like to share with you today…”
The real Ethan Mollick, who is attending the call, stops the video. “That is an entirely fake ‘me’,” he says. He explains that the video lecture was created by the AI program known as ChatGPT, in only seven minutes and at a cost of less than two dollars.
The AI had made the false recording with little effort by Mollick: he had instructed it to create a script that would “sound like Ethan Mollick talking about something.” He supplied it with a one-minute clip of himself discussing a random subject (gourmet cheeses) and a single photograph.
“Everything I’m showing you is commercially available technology that everyone has access to today,” he said, adding, “The world’s getting very weird, very quickly.”
What does the new AI technology mean for the business world, and specifically for retail? Mollick discussed all that and more.
Ideas for Marketing Collaborations—Within Seconds
Mollick further demonstrated AI’s capabilities—in real time—by posing a retail marketing question suggested by Baker managing director Mina Fader: brainstorm a hypothetical marketing partnership between Shake Shack and Louis Vuitton. Mollick typed into the AI program, “I am the CMO of Shake Shack and I want 20 ideas of ways we could collaborate.”
Within seconds, the concepts tumbled down the screen: A Shake Shack–Louis Vuitton photo booth. A themed event such as a fashion show or food festival. A branded food truck tour that would serve up Shake Shack favorites at select Louis Vuitton stores or events. An employee uniform collaboration. Limited-edition watches or jewelry. And on and on.
The AI could handle the nuts-and-bolts, too. At Mollick’s request, the system provided step-by-step instructions on how to implement particular ideas. It even generated the text of an email that could be sent to Shake Shack employees explaining their responsibilities in the new collaboration. Moreover, the program could readily change the email’s tone as directed: it could be written to sound straightforward, corporate-sounding, playful, or even be in rhyme.
Mollick then demonstrated that with more advanced AI systems, it was possible to generate personas for potential buyers and project how these customers would react to the collaboration. “It’s going to do our consulting work for us here, but it’s obviously making this up,” he noted. What might the hypothetical Anna, a 25-year-old fashion blogger, or Ben, a 35-year-old lawyer in a busy downtown office, or Chloe, a 19-year-old art student—think of the Shake Shack ankle boots, or the Louis-Vuitton branded food trays?
To top it off, the AI created an original photo shoot for the campaign, complete with models, designer handbags, and backgrounds.
The Effect on White-Collar Work
Despite all this, Mollick advised, don’t plan to fire your marketing department. He described AI as a useful tool, not a replacement for humans. “This is a force multiplier for experts,” he said. “There’s a lot of momentum in producing material that it can do for you. Then you check over the information, correct it, or plug in data and ask it to work from that.”
He cited early findings from research on productivity impacts. One found a 30%-60% increase in productivity among people doing marketing writing and other kinds of business writing. Their finished products were of higher quality, and the individuals also reported greater job satisfaction. “I’m doing some work with companies I can’t name right now, but I’m seeing up to 80% performance improvements in white-collar analytical work,” said Mollick.
AI’s Terrifying Side
Using AI comes with some big caveats, Mollick said. For one, “It makes stuff up all the time. It doesn’t know truth.” Some of us may have seen news items about how, for example, an AI-generated resume or biography will sometimes introduce completely fake credentials out of nowhere. Mollick recounted a recent experience with AI in which it decided to write a New Yorker magazine story that never existed. “It was guessing off the fact that I’d pasted in the URL for the New Yorker and gave the title of the piece, and it just made it all up like a college freshman who was BS’ing a whole bunch of plausible information.”
The dark side gets much darker. Mollick pointed out that these programs have been developing capabilities that no one expected. For example, how did the first version of these systems—a next-word-prediction algorithm—suddenly learn to do complex mathematics and play chess? What does it mean that not even the creators of AI programs themselves fully understand how these advances occurred?
Moreover, the ease with which fake video and audio can be created has significant implications. Consider the possibility of receiving a desperate phone call from a relative stranded overseas, asking for money, in a voice indistinguishable from your loved one. AI could be used for any number of nefarious purposes, such as crime, the bolstering of conspiracy theories with fake “evidence,” and attacks on other nations. Government regulation of AI is absent, and its creators are largely AI evangelists reluctant to rein in development of this revolutionary, profitable technology.
For Mollick’s part, he says he’s taking a pragmatic view. “AI programs are here, and we can talk all we want about how we wish they weren’t here, or wish they were regulated, or wish people were slowing down. None of that’s going to happen. So how do we use the positive side while being aware that this negative stuff is about to occur?”
A Competitive Advantage
Since the AI landscape is changing so fast, asked one board member, would it be smart to wait a few months and let things shake out before attempting business implementation? Mollick disagreed with this approach. “Here’s the trap everyone’s in right now: these systems are flawed, but they also increase performance hugely… And if you wait a long time for them to operate better, you’re losing your competitive advantage to everybody else who is both using them and learning how to use them.”
Mollick added that he believes retail and marketing will get hit “hardest and soonest” compared to other types of work, which is another reason to move quickly. He encouraged the board members to start actively experimenting with publicly available AI programs. Bottom line, he said, “If your team is not bringing AI to the table at every meeting you have, you’re missing out.”