Retailers: Here’s what you should be asking about AI
Image by Andy Kelly via Unsplash.
Artificial intelligence is developing at breakneck speed, and its applications in the retail industry across marketing, supply chain and store operations are increasing every day.
Questions about how best to use this technology in an ethical fashion, therefore, must be considered just as urgently—and they must be asked as early as possible in the design process of every AI-based solution.
That was the consensus formed during a panel on the ethics of AI in retail at NRF 2018: Retail’s Big Show between Anindya Ghose, The Venerable Tenzin Priyadarshi, Francesca Rossi, and moderator Daniel Hodges on Monday.
Priyadarshi, the director of the Ethics Initiative at MIT Media Lab, argued that asking hard questions about AI when framing algorithms and creating user experiences should not simply be seen as a matter of regulatory compliance for retailers, but rather as an approach to optimization—a way to improve products and help design them to “help human society flourish.”
“From the very beginning, it’s important to think of certain ethical parameters not in terms of restraint, but in terms of how you could do this better, how you could design this better,” Priyadarshi said.
What kind of ethical quandaries can AI present? Ghose, the Heinz Riehl Chair Professor of Business at the NYU Stern School of Business, said one of his chief concerns is the risk of building accidental or intentional discrimination into AI algorithms. While AI systems can be trained to process vast amounts of information, they can prioritize certain data sets over others in ways that can be either harmful or beneficial to certain individuals. Figuring out the root of such biases, Ghose said, is often “an incredibly difficult problem.”
“Think of AI as a hammer. You can use a hammer to break a house or you can use it to make a house. AI—and the toolkit associated with AI—can be used either to identify discrimination or it can be used to discriminate,” he said.
To a certain extent, Priyadarshi said, the issues AI pose are just like those presented by all new technologies: To what extent is it responsible for promoting a consumer’s well-being? How will it impact foundational social institutions? And will it be a vehicle of egalitarianism or elitism?
Rossi, the AI Ethics Global Leader and Distinguished Research Staff Member at IBM Research, said that companies that, in considering those questions, optimize their AI systems to be fair, transparent and explanatory will ultimately win favor with customers, who “like to understand what’s happening behind the scenes of the user experience.”
Taking the time to craft exemplary and ethical AI solutions can, Rossi said, slow down product development, but it’s a worthy investment that’s certain to pay off down the road.
“I think companies should know and be aware that there is competitive advantage and business value in actually forgetting about the short term benefits and thinking about these ethical issues for the long-term,” Rossi said.