Is AI adoption moving too fast? – RetailWire

Mar 23, 2021

According to KPMG’s “Thriving in an AI World” study, 53 percent of business leaders in retail said COVID-19 increased their pace of AI adoption, yet 49 percent believe adoption is moving faster than it should in their industry.

The broader survey of 950 business and/or IT decision makers found similar sentiments for other industries, including industrial manufacturing, 55 percent; and tech, 49 percent. The concerns were traced to debates surrounding the ethics, governance and regulation of AI.

Among retail respondents, 78 percent said it is difficult to stay on top of the constantly evolving AI landscape — a sentiment higher than leaders in other categories. Cybersecurity breaches (47 percent) and possible AI bias (45 percent) were found to be the top two greatest potential risks of AI adoption. Eighty-seven percent believe the government has a role to play in regulating AI technology.

Matt Kramer, national consumer and retail sector leader at KPMG, said in a statement, “The concern about the speed of adoption raises a caution flag for retailers, reminding them to ensure proper process and controls along with change management and effective training are put in motion to address the AI adoption risks.”

Among the encouraging responses from retailers:

• Eighty-one percent said AI is moderately to fully functional in their enterprise, up 29 percentage points from a year ago. 
• Seventy-eight percent said their company has an AI code of ethics, up 34 percentage points;
• Ninety percent indicated that their employees are prepared and have the skills for AI adoption, up 47 percentage points.

A recent survey from Rackspace Technology of 1,870 organizations across industries showed only 17 percent of companies have mature AI/machine learning capabilities, but other adoption challenges were cited.

The top five barriers to actionable insights were data quality issues, cited by 31 percent; followed by lacking the capabilities to effectively manage our data, 30 percent; data being stored in too many different systems, also 30 percent; inability to process data quickly enough to act on it in time, 29 percent; and difficulty collating, structuring and integrating data in a meaningful way, 28 percent.