As artificial intelligence capabilities proceed to increase, many executives are grappling with a vital determination: Ought to they put money into upskilling their existing teams or herald exterior consultants to jumpstart their AI initiatives?
With the worldwide AI consulting market projected to succeed in $72.5 billion by 2025, with a compound annual progress fee of 40.3% from 2020 to 2027, and 63% of enterprises planning to increase their AI investments within the coming yr, the demand for exterior experience is clearly on the rise. However that does not imply there is a one-size-fits-all resolution, in keeping with Hakan Kardes, chief expertise officer at Alignment Well being.
“There are plain benefits to bringing in exterior consultants, particularly early within the AI journey,” he mentioned, including that exterior consultants provide rapid entry to specialised expertise and information, that are key for accelerating essential initiatives and exploring new applied sciences.
However Kardes says the problem lies in “making certain the exterior experience is successfully built-in with the inner group…to stop long-term dependency.” And whereas upskilling has the advantage of embedding AI experience deep inside a corporation for the long run, this path requires vital time, assets, and dedication, which not each firm can afford, he says.
Stephen Boyer, co-founder and chief innovation officer at cyber threat administration agency Bitsight, says “The straightforward factor to do is to rent PwC as a result of they have individuals who’ve been doing this for a decade, however on this case no one’s been doing this factor for a decade. Folks have been doing machine studying and different components of AI for a very long time, however [generative AI] is considerably novel.”
“Hiring expertise is tremendous costly,” he mentioned. “We did not see the expertise we would have liked out there. Except you had been an AI guru at Meta or Google, you were not a part of that foundational stuff. However we knew we would have liked to deal with this.”
As an alternative of hiring exterior AI consultants, he developed a “tiger group” — a specialised inside group devoted to experimenting with AI.
“We got down to do one thing that we might name a fast win,” Boyer mentioned. “It was the popularity that we did not have these expertise however we’re going to put money into doing it … with out creating a whole division and hiring lots of people. So we put two engineers on it and my time and mentioned, ‘Let’s go study, construct these relationships, after which do these experiments and see if we are able to ship.'”
Gen AI outcomes
Bitsight’s AI experiment centered on automating a labor-intensive job: leveraging generative AI to course of huge quantities of cybersecurity articles produced every day throughout a number of languages. The Bitsight group was capable of automate the method, educating AI to learn, analyze, and determine key particulars from articles with accuracy. “We handed it any article in nearly any language and had been capable of decide whether or not it was a couple of breach and who was implicated,” says Boyer.
Based on Boyer, Bitsight’s method yielded spectacular outcomes: “We acquired it to be extremely dependable — about 90% of the time it agreed with what a human would say,” he mentioned. Though solely a yr in, Boyer says the group realized “what labored, what did not, and what you could possibly depend on.”
Now he is pushing that functionality out to the remainder of the group by embedding the people who find themselves doing the experiments into the groups which can be doing the event, and Boyer says this experiment has made the corporate extra assured and higher geared up to develop a roadmap for future AI initiatives.
Past simply the outcomes, Boyer highlighted the joy his group skilled as they dove into AI experimentation. “Our engineers had been energized by the problem,” he mentioned. “It was cool to see them studying one thing new and bringing their creativity to the challenge. Seeing them really feel empowered by this chance was simply as rewarding because the technical outcomes.”
He emphasised, nevertheless, that governance and threat administration are essential elements of the technique to keep up management over information and AI fashions. “We needed to verify we had the fitting entry controls, that none of our information was being leaked, and that we weren’t inadvertently exposing ourselves to threat,” Boyer mentioned.
This cautious method helped handle dangers resembling hallucinations — the place the AI generates info that seems appropriate however is inaccurate. “We have realized that AI is nice for sure duties, like summarizing paperwork, however we needed to sandbox it to verify it did not go off the rails,” he mentioned.
Like Boyer, Kardes opted to put money into Alignment Well being’s inside experience. The medical health insurance firm has been dedicated to constructing its AI capabilities over time, beginning with the event of a proprietary information expertise platform, AVA, a decade in the past. Right now the corporate has greater than 200 AI fashions embedded in varied purposes and workflows, driving improvements in senior care.
A hybrid work method
Whereas Bitsight and Alignment Well being are lucky to have a powerful inside technical basis, not each firm has the posh of upskilling from inside.
Based on Ryan Sutton, government director of expertise follow at recruiting agency Robert Half, a latest survey discovered that 90% of expertise leaders plan to implement AI initiatives this yr, but 41% of them cite a scarcity of employees with AI expertise and experience.
“Tech leaders for corporations of all sizes recognized AI and machine studying as areas with probably the most evident expertise gaps, together with information science and expertise course of automation,” he mentioned.
Given this panorama, Sutton advises corporations to undertake a hybrid method — one that enhances their everlasting employees. “It is essential to make sure that consultants not solely present the precise expertise or expertise required but additionally improve and assist the general well being of your workflow,” he added.
