artificial intelligence 
Can artificial intelligence make us more creative and innovative? It’s a subject of hot debate and discussion. A recent analysis out of the Gottlieb Duttweiler Institute suggests that, yes, AI can help us expand our range of innovation.
Along with handling the mundane, “AI can also take over more creative tasks by identifying patterns in data that humans would not have found,” the study’s author, Jan Bieser, points out. “In this case, AI does not just take over tasks that would be time-consuming; it might provide insights humans would have never found themselves.”
There’s only one rub: how viable is the data running through these AI systems? AI doesn’t appear in a vacuum. It’s the result of the data behind it. Many industry experts are concerned that companies aren’t paying enough attention to the data that is driving their decision systems, data which may be deficient, too limited in scope, or stale. Dry data dries up innovation as well. “Your data is constantly evolving as circumstances shift rapidly,” says Arijit Sengupta, CEO and founder of Aible. “Many AI projects fail because they are run on outdated or useless data and ignore the business realities.”
Data may be useless, or there simply may not be enough of the right data. “The most common mistake businesses make when implementing AI is believing that all of the necessary data exists in closed-loop systems,” says Melanie Nuce, senior VP of innovation at GS1-US, a nonprofit consortium developing digital trading standards. “Businesses may deploy AI with the belief that they can find value from the technology using all of their own data, but for AI to scale effectively, the data will likely need to be ingested and shared across trading partners.”
As reliance on AI grows, there is a risk of decisions going astray due to underlying data issues. “A mistake even the most established enterprises continue to make is relying on data as the sole source of truth,” says Sengupta. “We need to understand that traditional AI doesn’t have any understanding of your goals, cost-benefit tradeoffs or capacity constraints. All it knows is what is in your data. For that reason, data alone is the wrong basis for a successful AI strategy.”
Poor data is the reason many AI implementations don’t deliver. “Biased or insufficient data can have serious long-term consequences for any AI project,” says Shalabh Singhal, CEO of Trademo. “Most companies complain of poor ROI even after spending most of their budget on data collection. What they fail to understand is the importance of collecting the right data and further, cleaning and labeling it.”
To see the full benefits of AI adoption, “feed it complete, accurate and consistent data,” says Nuce. “When data is not structured or harmonized, business processes cannot be automated, and the investment is wasted — along with valuable time and resources. The insights we gain from AI are only as strong and accurate as the data that feeds it.” She calls for stronger industry standards to “ensure the right data is being collected in machine-readable ways, so companies can achieve value faster.”
With data standardization, companies will be able to innovate at a faster pace, Nuce continues. “Access to greater amounts of high-quality data enables data scientists to build algorithms that function at a much quicker learning capacity and require less supervision and management. We are still discovering what AI can do for mainstream businesses, but with external collaboration and data sharing, the possibilities are endless.”
When designing AI-driven processes, “start with the end in mind,” says Arijit Sengupta. “When you start with a hammer, everything looks like a nail. That’s the first, and sometimes fatal, mistake. The available data may simply not support that use case, and AI can’t do anything if the data is not available.”
It boils down to not implementing AI for AI’s sake. The most effective AI projects are “business objective first,” Sengupta continues. “If you want to increase revenue, start by better targeting your sales efforts, improving your marketing strategy, reducing customer churn, or increasing partner sales. The right approach points the AI at all the available data and figures out which use cases can be supported by the data to improve the business objective.”
Follow me on Twitter.
