Long thought of as a tool for improving operations, AI is beginning to play a role earlier in the enterprise process, closer to product design.
Using the analytics capabilities of AI to design new products and services is a promising frontier in enterprise IT, according to Marshall Van Alstyne, Boston University Questrom School of Business professor, who spoke at the MIT Sloan CIO Symposium on Monday.
By integrating internal data silos and drawing on external data sources, companies can create dynamic AI ecosystems, spurring innovation, improving efficiency, and creating new business opportunities.
Most organizations moving AI closer to product design are digitally native, but some legacy companies are now also following this model, said Thomas Davenport, professor of information technology and management at Babson College.
Targeted advertising, customer service automation, quality control, and logistics are all areas where AI has proven its business utility. When effectively paired with big data, AI can reshape the way companies develop products.
To keep up with Uber, eBay, and Facebook, which are already using AI at scale, legacy companies like Airbus, Anthem, and DBS Bank are working AI into their systems.
“It’s like a Disneyland of data in this environment,” Davenport said. “That can improve your AI models, which lets you improve your services, which reduces customer friction, which helps you acquire more customers, which gives you more data.”
This virtuous circle has created entirely new marketplaces, primarily through the ML-powered search functions that enable e-commerce, using data to match buyers with sellers on sites like Amazon and eBay. While AI recommendation systems remain a powerful business tool, new ways of thinking about AI technologies are gaining ground.
Beyond recommendations, into design
Netflix is one example of a business built around an ML recommendation algorithm. The company began coupling AI with growing data assets to create streaming content to help guide the development process, most notably in the creation of the critically acclaimed show “House of Cards.”
Amazon similarly relied on AI for “The Marvelous Mrs. Maisel,” Van Alstyne said.
“A recommender system takes existing products and tells you what will be successful. But you can also use a recommender system to create products that don’t yet exist. So, you can reverse and invert it,” Van Alstyne said.
For companies that commit to pairing AI with cloud-based APIs, the gains in revenue and market cap can be significant. A 2021 study Van Alstyne co-authored, which analyzed 200 firms that adopted APIs, found those that let AI learn from third-party interactions saw a 4% gain in market cap over two years and 38% over 16 years.
But there's a cyber risk catch. “When you externalize the APIs, there’s a 1.2% chance of getting hacked each month. You are going to get the huge gains, but it comes with something you have to manage,” Van Alstyne said.
Companies that can’t afford that level of risk exposure can still benefit, according to Vipin Gopal, Eli Lilly & Company chief data and analytics officer. Gopal sees opportunities within the pharmaceutical industry for large organizations to make major gains simply by sharing currently siloed data.
“It's something that we see across the industry," said Gopal. "Various parts of the organization grew up in silos — the research side, the clinical development side, the packaging and commercial side. There is a tremendous amount of data that is resident in various parts of the organization that are largely disconnected.”
