AI is forcing many companies to make difficult, costly hiring decisions

• As artificial intelligence capabilities continue to expand, many executives are grappling with a crucial decision: Should they invest in upskilling their existing teams or bring in outside experts to jumpstart their AI initiatives?
• "There are undeniable advantages to bringing in external consultants, especially early in the AI journey," says Hakan Kardes, chief experience officer at Alignment Health.
• A recent survey by Robert Half discovered that 90% of technology leaders plan to implement AI initiatives this year, yet 41% of them cite a lack of staff with AI skills and expertise.

As artificial intelligence capabilities continue to expand, many executives are grappling with a crucial decision: Should they invest in upskilling their existing teams or bring in outside experts to jumpstart their AI initiatives?

With the global AI consulting market projected to reach $72.5 billion by 2025, with a compound annual growth rate of 40.3% from 2020 to 2027, and 63% of enterprises planning to increase their AI investments in the coming year, the demand for external expertise is clearly on the rise. But that doesn't mean there's a one-size-fits-all solution, according to Hakan Kardes, chief experience officer at Alignment Health.

"There are undeniable advantages to bringing in external consultants, especially early in the AI journey," he said, adding that external consultants offer immediate access to specialized skills and knowledge, which are key for accelerating critical projects and exploring new technologies.

But Kardes says the challenge lies in "ensuring the external expertise is effectively integrated with the internal team…to prevent long-term dependency." And while upskilling has the benefit of embedding AI expertise deep within an organization for the long term, this path requires significant time, resources, and commitment, which not every company can afford, he says.

Stephen Boyer, co-founder and chief innovation officer at cyber risk management firm Bitsight, says "The easy thing to do is to hire PwC because they've got people who've been doing this for a decade, but in this case nobody's been doing this thing for a decade. People have been doing machine learning and other parts of AI for a long time, but [generative AI] is somewhat novel."

"Hiring talent is super expensive," he said. "We didn't see the experience we needed in the market. Unless you were an AI guru at Meta or Google, you weren't part of that foundational stuff. But we knew we needed to address this."

Instead of hiring external AI consultants, he developed a "tiger team" — a specialized internal team dedicated to experimenting with AI.

"We set out to do something that we would call a quick win," Boyer said. "It was the recognition that we didn't have those skills but we are going to invest in doing it … without creating an entire division and hiring a lot of people. So we put two engineers on it and my time and said, 'Let's go learn, build these relationships, and then do these experiments and see if we can deliver.'"

Gen AI results

Bitsight's AI experiment focused on automating a labor-intensive task: leveraging generative AI to process vast amounts of cybersecurity articles produced daily across multiple languages. The Bitsight team was able to automate the process, teaching AI to read, analyze, and identify key details from articles with accuracy. "We handed it any article in almost any language and were able to determine whether it was about a breach and who was implicated," says Boyer.

According to Boyer, Bitsight's approach yielded impressive results: "We got it to be highly reliable — about 90% of the time it agreed with what a human would say," he said. Although only a year in, Boyer says the team learned "what worked, what didn't, and what you could rely on."

Now he's pushing that capability out to the rest of the organization by embedding the people who are doing the experiments into the teams that are doing the development, and Boyer says this experiment has made the company more confident and better equipped to develop a roadmap for future AI initiatives.

Beyond just the results, Boyer highlighted the excitement his team experienced as they dove into AI experimentation. "Our engineers were energized by the challenge," he said. "It was cool to see them learning something new and bringing their creativity to the project. Seeing them feel empowered by this opportunity was just as rewarding as the technical outcomes."

He emphasized, however, that governance and risk management are critical components of the strategy to maintain control over data and AI models. "We wanted to make sure we had the right access controls, that none of our data was being leaked, and that we weren't inadvertently exposing ourselves to risk," Boyer said.

This cautious approach helped manage risks such as hallucinations — where the AI generates information that appears correct but is inaccurate. "We've learned that AI is great for certain tasks, like summarizing documents, but we had to sandbox it to make sure it didn't go off the rails," he said.

Like Boyer, Kardes opted to invest in Alignment Health's internal expertise. The health insurance company has been committed to building its AI capabilities over the years, starting with the development of a proprietary data technology platform, AVA, a decade ago. Today the company has more than 200 AI models embedded in various applications and workflows, driving innovations in senior care.

A hybrid work approach

While Bitsight and Alignment Health are fortunate to have a strong internal technical foundation, not every company has the luxury of upskilling from within.

According to Ryan Sutton, executive director of technology practice at recruiting firm Robert Half, a recent survey found that 90% of technology leaders plan to implement AI initiatives this year, yet 41% of them cite a lack of staff with AI skills and expertise.

"Tech leaders for companies of all sizes identified AI and machine learning as areas with the most evident skills gaps, along with data science and technology process automation," he said.

Given this landscape, Sutton advises companies to adopt a hybrid approach — one that complements their permanent staff. "It's important to ensure that consultants not only provide the specific skills or experience required but also enhance and support the overall health of your workflow," he added.