Has COVID-19 Had An Effect On Enterprise AI Adoption?

The coronavirus pandemic has brought many industries to a halt, with businesses reducing hours, sacking employees, and halting new projects.

Artificial intelligence projects could definitely fall into the dispensable category, but that’s not the case according to a recent report from FICO and the market intelligence firm Corinium.

The Building AI-Driven Enterprises in a Disrupted Environment report surveyed more than 100 c-level data and analytic executives and conducted in-depth interviews to understand how organizations are developing and deploying AI capabilities. It found AI demand has risen during the pandemic for a majority of businesses surveyed.

SEE ALSO: How AI is Changing the Healthcare Industry

That said, there have been changes in priority, as business leaders look for more short-term successes.

“I actually think we may see AI and ML adoption slow down a bit in certain areas, and people will select ‘safer’ technologies and focus on responsible AI in the very near-term, with respect to COVID-19,” said Scott Zoldi, CAO at FICO.

How short-term success is defined will depend on the original goals of the AI project. Customer satisfaction, risks mitigated, revenue, or saving generated or process optimization are key metrics normally used, although some of those may take time to materialize.

Many businesses on the periphery of developing an AI model still find skill shortages to be a major barrier to deployment and adoption. 65 percent surveyed by FICO regarded it as the highest barrier, with legacy system integration and regulatory compliance also considered barriers.

Interestingly, financial services, manufacturing, and advertising industries were the hardest pressed to find the right AI talent, while pharmaceutical, insurance, and retail struggled with legacy system integration.

Some barriers are becoming less of an issue, according to FICO technical leaders are receiving more support from board and C-level executives, and do not have to constantly prove ROI for projects.

This, according to a Forrester analyst Mike Gualtieri, is a sign of the AI and ML market maturing. The fact that 63 percent of those surveyed said they are scaling their AI capabilities is another sign of maturity.

The amount of AI adoption is heavily dependent on the industry’s IT sophistication. As Bart Pietruszka, head of analytics at HSBC said to FICO, industries that are process-heavy and people reliant are “not easily able to detach themselves from the classical way of doing things.”

The best way to ensure organizations use AI effectively, according to those surveyed, is by having a streamlined development-to-product environment. However, 67 percent said they do not monitor models for accuracy drift, meaning that the majority of executives surveyed stop paying attention to the model post-deployment.

In the current pandemic, AI models will be experiencing unique levels of traffic and interest, so it is imperative businesses retain talent to monitor and retrain models.

“Models are probabilistic because they’re trained on historical data, when the environment changes it needs to be re-trained,” said Gualtieri in a webinar. “When they are in production they need to be monitored for drift, to make sure that they’re making good decisions.”