AI has become a vital component to how businesses operate during the COVID-19 pandemic, but as adoption increases, IT leaders are now faced with the problem of finding workers with the right AI skills.
"Many of us have seen our digital lives change," said Scott Zoldi, chief analytics officer at credit scoring and analytics company FICO. "We're going to become more and more dependent upon digital, but now I think it's going to take much more speed."
In FICO's recent report, Building AI-Driven Enterprises in a Disrupted Environment, which surveyed 104 chief data and analytics officers, 57% of respondents said demand for AI products has risen as a result of the COVID-19 pandemic. This has made figuring out how to hire data scientists an even bigger priority than before.
Zoldi said he's seeing consumers making bigger decisions digitally than ever before, which is putting pressure on enterprises to use AI to facilitate more digital transactions.
This article is part of
"We're going to see more use cases around having to verify customers for fraud, authenticate customers, originate new types of business loans and/or personal loans," he said. As a result, organizations will have more needs to find out the data provenance "and then new models to combine all the information into a decision around what that sort of loan would look like or what that sort of decision would be," he added.
AI skills needed in the enterprise
Mark Chamberlain, vice president of product development at ADP, is also seeing AI playing a key role during the pandemic and expects being able to utilize it to automatically gain better insight based on data to be key even after.
"Data cloud right now is one of our bigger growth areas at ADP," he said. "We use [our data] to provide insights and embed that in products that we sell back to our customers."
ADP's AI efforts haven't slackened during the pandemic, and Chamberlain said he's bullish about continued forward progress in this area even after the pandemic.
To take advantage of these AI trends, Chamberlain said, IT leaders are trying to figure out how to hire data scientists with diverse sets of experiences.
"Somebody that knows server architectures, operating systems, networking or networking hardware, storage arrays and backup [and] databases like an Oracle DBA or MySQL DBA," Chamberlain said. And because Python applies to a lot of the infrastructure, there's a demand for software engineers. "There's a lot of tools like Ansible that will allow you to do a lot of orchestration of that automation that you've done to support, maintain and update large-scale infrastructure."
According to Zoldi, organizations are also searching for skill sets that include program and project managers. "We're looking for someone to enforce the standards of work, audit trails, Agile development processes and the scrum processes."
He said it isn't just about finding brilliant data scientists, it's also about finding people that may have been in the trenches and have a good business and process mindset.
"[There's] a lot of process around the development of these [AI] models and I think that's one key thing -- being enamored with the latest algorithm is not so much what they're hiring for, it's around really having this sort of process around how we develop," he said.
Finding AI skills poses as a challenge
The downside is, as organizations look to AI tools to support their businesses during the pandemic, many are finding it hard to acquire the workers and the AI skills needed to support the technologies. "A clear example that I've seen is many clients want to expand [their] data science but were in a hiring freeze," said Peter Krensky, an analyst at Gartner.
"We have a lot of data and we're starting to do more and more with it," Chamberlain said. "The challenge is finding the skill. Having business analysts [and] engineers or subject matter experts to understand how to go in and interpret and create these models that will make sense of the data."
Zoldi said one main reason it's difficult to find data science talent is organizational.
In the FICO report, 65% said building a team with the right skills is a large or medium barrier to AI adoption. According to Zoldi, this is in large part due to not having all executive teams involved in the decision-making process, having difficulty enforcing standards across several teams and not having analytics under one leader, something FICO has addressed.
"We have the analytic structure, which is tremendously important to addressing the ethical AI approaches to adopting AI and machine learning, and an opportunity there," he said. "I think the right team doesn't necessarily just mean data scientists and that's the other aspect of what's behind some of that equation too."
There's also the ongoing problem of trying to fill the skills gap. "There's a talent gap in the space," Krensky said. "[And] this talent moves a lot. Data scientists have infamously short tenures because they're always getting recruited and they have the rare leverage in a job where they can get either an equal or better offer somewhere else."
Reskilling employees internally
But acquiring the necessary AI skills isn't just about hiring data scientists. There are IT leaders who have switched gears and are taking advantage of these unprecedented times to provide employees with the opportunity to learn and be trained on a new technology. "Reskilling is a great thing right now because it gives your associates time to stop thinking about the day-to-day [tasks] and it gives them something new to think about," Chamberlain said.
This is something ADP has focused on in its infrastructure and operations team, reskilling over 2,000 associates. "We got them away from that traditional hardware mentality [where] they follow a playbook and follow steps and type commands," he said. "We reskilled [them] to be able to program."
And the logical step was to introduce them to Python because it's not a compiled language, according to Chamberlain. "It's an interpreted language just like shell scripting, but it gives you a lot of flexibility around being able to create reusable components," he stated. "Python is very popular right now, so we transitioned and upscaled associates to do that. And that helps with our automation that we're doing and engineering to get rid of the manual as well."
And by reskilling these employees, they've also gained a common language in the workspace. "When you have a network engineer, a server engineer and a database engineer in a room together, they don't talk a common language," Chamberlain said. "They're talking very specific in the areas they focus on -- but being able to reskill them and using Python as a common language, they can start talking similar terms that will allow them to gain efficiencies over time. And that was really core to what we did with our transformation."