Dive Brief:
- To locate where AI can help companies gain efficiency, organizations should identify problem areas and the data that can power AI systems, according to AI author and entrepreneur Tom Taulli, speaking Tuesday at Laserfiche Empower 2021.
- Organizations that "start off the project saying, 'let's do something cool,'" are in "big, big trouble," said Taulli. Instead, companies need to direct their AI efforts toward solving tedious problems within critical company processes, Taulli said.
- AI projects frequently fail when organizations rush to employ data scientists and invest in AI capabilities without first confirming they have access to the data that can help them improve on a specific pain point, according to Taulli.
Dive Insight:
Enterprise need for deploying bleeding-edge tech is a clear trend in the year ahead. In January, job posts for emerging technologies drove 28% of new IT job growth.
But business leaders often struggle to locate the use cases that have the highest impact on efficiency.
Less time spent on repetitive tasks means employees focus on making the company more competitive and offer better service to customers, according to Taulli. As a byproduct, businesses expect employee experience to improve as they deploy with more automation capabilities.
Technology can also let businesses separate applications from specific use cases. H&M devoted its machine learning efforts last year to create reusable components, available to the entire organization through component libraries.
To determine if an AI platform is delivering on its objectives, Taulli recommends establishing a center of excellence, which consists of a group of people who can take ownership over a project, track its progress and monitor performance over time.
But to successfully monitor AI, companies will need access to specialized talent, a top hurdle toward successful AI implementation cited by 37% of IT leaders in an IBM survey.
"You just don't throw AI out in the wild and let it roam free," said Taulli. "You want humans in the loop, you want some oversight, you want some governance."
