Robots in retail are no longer the stuff of science fiction. The cluster of technologies we call Artificial Intelligence is still new and, some would say, in its infancy. But over the past decade, terms such as “machine learning” and “predictive analytics” have evolved from arcane jargon to business buzzwords.

Across all retail segments, AI has stepped out from R&D labs and started to transform every aspect of the way businesses engage and serve customers. From the sales floor to the head office to the warehouse, AI is harnessing the power of data and automation to enhance the customer experience, increase efficiency, and boost profits.

And the tech is no longer just for giant players, like Lowes with its roaming LoweBots and Amazon with its cashierless Amazon Go store. A growing number of vendors now provide off-the-shelf solutions, many of them cloud-based, that put AI applications within reach of any retailer.

The challenge is no longer to imagine the possible uses for AI. We now have myriad examples of how AI can take both customer service and company business models to the next level.

The new frontier for AI-enabled retail is risk mitigation.

As retailers become “smarter,” they also become more vulnerable to a variety of technology-related threats. Businesses need to be thoughtful when using AI because the ability to recognize and adapt to risks makes the difference between a technology success or a technology failure.

What Can’t AI Do?

Retailers are now using AI to:

• Install “smart shelves” that automatically track the supply of merchandise and create an alert when it’s time to restock.
• Generate heat maps showing customer traffic through a store
• Tailor in-store digital advertising to consumers based on their real-time shopping behavior
• Enable customers to check out items without touching a POS device
• Advise customers on which products to buy and where to find them
• Track the availability of in-store promotional materials

As the hardware required for advanced computing becomes smaller and more affordable, it seems to open boundless opportunities for inventive retailers. At the same time, it’s important to implement AI solutions carefully because of the potential risks, which can impact customers and businesses.

AI can capture, compute, and categorize many different forms of data. But it can’t yet apply ethical reasoning to complex situations. That’s where human intelligence needs to come into play.

Seven Risks of Adopting AI Technology

Dr. Jason Rhinelander is an Assistant Professor of Engineering and Head of Technology and Innovation for the David Sobey Centre at Saint Mary’s University. He conducts research to improve the efficiency of AI and computer vision, focusing on computing systems that are embedded in the retail environment.

While Dr. Rhinelander’s research addresses the technical challenges of AI, he sees the business risks of emerging technologies as presenting the biggest hurdle to retailers. From his perspective, retail leaders should consider the following seven risks as they integrate AI into their operations:

Risk #1: Putting technology ahead of the business problem and business strategy. With so many AI options now readily available, it’s easy to be seduced by apparent potential. Don’t let shiny technology distract you from the discipline of building a solid business case. If you introduce an AI “solution” without linking it directly to measurable outcomes and your overall business strategy, then you’re probably just creating a problem.

Risk #2: Jeopardizing customer privacy. Dr. Rhinelander recommends one simple question to test the level of data security an AI application provides: “Will your customers feel like family?”

Protecting customer privacy should be your number one concern. Says Dr. Rhinelander, “Customer data should be at the centre of the solution.”

If you’re considering a cloud-based application, make sure you know where the provider’s servers are located as privacy laws vary from country to country. Also take some time to research the provider. Have they appeared in the news in connection with any privacy issues?

Risk #3. Making decisions based on biased data. Any AI solution is only as good as the data set it has learned and uses to operate. Yet many data sets are inherently biased.

Ask solution providers about where their data sets come from and how they’re scrutinizing their data to make it as bias-free as possible. Ethical companies will be more than happy to answer your questions.

Risk #4. Failing to comply with changing regulations. Regulations concerning data and AI are evolving almost as rapidly as the technologies they govern. Canada may soon introduce some of the most stringent AI regulations in the world, and other jurisdictions are considering similar legal frameworks.

If the Artificial Intelligence and Data Act (AIDA) that’s currently before Parliament becomes law, fines for noncompliance could add up to millions of dollars. To avoid such penalties, make sure you keep up to date with local, national, and international laws as they emerge.

Risk #5. Rushing into early adoption. The first-mover advantage isn’t always a benefit when it comes to experimenting with new technology, especially if you’re a small or medium-sized retailer.

Depending on the financial commitment and risks, the wisest bet for a small or medium-sized retailer may be a strategize-and-see approach. You may want to prioritize developing an innovation strategy,  and then wait until the technology matures before implementing the plan. Alternatively, small and medium-sized organizations can partner with a research organization, allowing innovation to proceed while minimizing risk.

Risk #6. Believing that ignorance is bliss. When an organization is considering introducing AI, everyone in the C-suite needs to get familiar with some technology basics. Don’t make the CIO solely responsible for understanding and evaluating new technologies. It’s risky to let most of the leadership term bask in willful ignorance. Instead, invest time in bringing all executives up to speed so they can make decisions from an informed point of view.

Risk #7. Creating experimental siloes. Successful innovation requires coordinated effort across an organization. When individual departments start investing in one-off AI solutions, without consulting one another, risks of various kinds compound. New technologies that should improve efficiency can drag down productivity when applications don’t interact smoothly or duplicate effort. Poorly governed AI projects, which don’t align with strategy at the organizational level, can drain finances, test team morale, and create ethical issues.

Integrating AI into a retail business requires thoughtful, centralized oversight, especially with regards to ethical concerns.

Success Depends on Sound Ethics

Dr. Rhinelander notes that researchers are working hard to make AI “more transparent.” This should make it easier for retail leaders to understand how an AI algorithm learns from data, makes its decisions, and functions internally. Such visibility into the inner workings of AI will help address some of the ethical issues that AI raises.

However, retailers also need to take preventative action at the organizational level. Dr. Rhinelander would like to see retailers appoint an “ethics overseer,” someone to serve as the interface between AI vendors and the C-suite. This person, says Dr. Rhinelander, should take the lead in developing an “ethics framework at the C-level” that applies throughout the organization.

The role of ethical guardian could be taken up by the Chief Innovation Officer, the Senior VP for Strategy & Innovation, or the VP for Transformation. The title doesn’t matter so much as the responsibility it entails.

The ideal candidate for this critical position brings to the senior boardroom the analytical mind and moral sensibilities of a philosopher or lawyer, along with deep technical knowledge. They’re also able to communicate clearly with colleagues across various business functions.

As AI moves into every corner of the retail world, brands large and small will need to guard against new and evolving risks. To successfully harness the power of AI, retailers will need to integrate technology and business strategy while operating from a position of ethical clarity.

Check out Part 1 of November's edition of the Retail Insights Series as Dr. Rhinelander discusses AI Applications in Retail: