Breaking News

CIOs face uphill climb in finding skilled artificial intelligence talent

For the past four years, the strongest demand for people with AI skills has not come from the IT department but from other business areas, Gartner data shows.

More for CXOs

New data from Gartner Inc. suggests that the recruiting, management, and retention of artificial intelligence talent (AI) will be a strategic challenge globally for the foreseeable future. For the past four years, Gartner found, the strongest demand for talent with AI skills has come from non-IT departments.

Departments recruiting AI talent in high volumes include marketing, sales, customer service, finance, and research and development, Gartner said in a press release. “These business units are using AI talent for customer churn modeling, customer profitability analysis, customer segmentation, cross-sell and upsell recommendations, demand planning, and risk management.”

See: Top 5 things to know about the state of artificial intelligence (TechRepublic)

Gartner TalentNeuron data released Wednesday shows the total AI jobs posted by non-IT departments in Top 12 countries by GDP, grew 74%, to 156,294 through March 2019, from 89,895 in July 2015. Jobs posted by IT departments surged in that time to 68,959 from 14,900, but the number of AI jobs posted by IT is still less than half of that stemming from other business units. 

Image: Gartner Inc

Peter Krensky, research director at Gartner, said the data echoed the existing trend of analytics activity, investment and talent moving from IT to business. “Many of the same challenges and points of friction from that transition are appearing again, but some leaders are successfully applying the lessons learned from the past decade,” he said.

Expanding and innovative university programs, online education, and improved tools will help address the situation, Krensky said, “but leaders should expect the talent gap to be a persistent issue.”

Krensky said in the press release that “high demand and tight labor markets have made candidates with AI skills highly competitive, but hiring techniques and strategies have not kept up.” 

He pointed to a 2019 Gartner AI and Machine Learning Development Strategies Study, in which respondents ranked “skills of staff” as the No. 1 challenge or barrier to the adoption of AI and machine learning (ML).

According to the 2019 Artificial Intelligence Index Report, released by Stanford University’s Human-Centered Artificial Intelligence Institute, in the US, the share of AI jobs grew from 0.3% in 2012 to 0.8% of total jobs in 2019. The report found that at the graduate level, AI has rapidly become the most popular specialization among computer science Ph.D. students in North America.

Gartner said in its findings, which will be discussed at its scheduled upcoming IT symposiums, that a significant portion of AI use cases are reported from asset-centric industries supporting projects such as predictive maintenance, workflow and production optimization, quality control and supply chain optimization. 

“AI talent is often hired directly into these departments with clear use cases in mind so that data scientists and others can learn the intricacies of the specific business area and remain close to the deployment and consumption of their work,” the report said.

Krensky said CIOs should encourage and invest in education for current staff wherever possible. “Many data engineers, citizen data scientists, machine learning engineers and even some expert data scientists evolved out of other job roles and educational backgrounds,” he said. “They should liaise with various business leaders to gain visibility into their current AI initiatives and staffing ambitions.”

What’s more, Krensky said, “They should explore formalized support relationships between IT and teams producing AI assets and create an enterprise strategy for AI governance.”

“Finally,” he added, “they should help other leaders keep an eye out for underdeveloped or poorly conceived AI projects negatively impacted by the talent deficit and help solve common problems (poor data quality, failure to detect bias, creating black boxes, overfitting, etc.).”

Also see

Image: Getty Images/iStockphoto