Dive Brief:

• Cost reductions, revenue growth and time savings are the top three criteria organizations use to measure the success of an AI or ML initiative, according to research from SambaNova Systems published Tuesday. The company surveyed 600 professionals at director level or higher. 
• Leaders also cited new product development and time to insight as markers of AI/ML project success.
• Organizations are finding a connection between AI strategies and the bottom line, as more than three-quarters of companies see AI/ML as very important for revenue generation, the report finds. 

Dive Insight:

Businesses turn to AI in search of improved operations, but leaders will need to move their organizations out of AI infancy and into more mature adoption stages, where the technology can truly revamp operations.

Mature businesses are deploying AI/ML first at a departmental level, allowing the technology to revamp and transform application workflow, said Marshall Choy, SVP, Product at SambaNova. 

But the route to AI maturity at scale isn't simple.

Data quality can set the pace for the development of applications, according to Deere CTO Jahmy Hindman. Manual data labeling can hold up the adoption process, so some companies are turning to automatic labeling. 

In certain industries such as healthcare or banking, customizing AI models is harder due to a complex data ecosystem. CVS built a personalization engine to increase prescription refill rates and reduce gaps in treatment — but first it needed to contend with the sparsity of customer data.

To make AI sustainable at scale, enterprise leaders must cultivate the right data to power models, put in place internal operating models and address the question of whether to buy or build technology internally, according to Deloitte. 

One strategy to strengthen the speed to enterprise-level AI adoption is to establish an AI center of excellence, said Choy. These units help adoption by taking on a formal role of governance across the organization, addressing the business, organizational and ethical dimensions of AI.