Highlights:

  • The lack of well-trained IT people is often cited as a hindrance to the widespread implementation of various information technology and Artificial Intelligence (AI) forms.
  • Only 3% of respondents named security the biggest obstacle, ranked above the inability to assess value, a lack of understanding of AI’s benefits and uses, and data access.

Businesses are starting to embrace automation and Artificial Intelligence (AI) even though there are still a few obstacles to be overcome. These obstacles may not be what many people think they are.

The lack of well-trained IT people is often cited as a hindrance to the widespread implementation of various information technology and AI forms. Gartner recently published a survey consisting of new data that seems to dispel concerns about a scarcity of AI professionals. Seventy-two percent of the businesses that participated in the survey said they had or could easily obtain the AI talent required.

While everyone is developing AI models, production is more difficult

While there’s no dearth of talent, transitioning from pilot to production is an issue. Gartner’s survey found that only a fraction of the AI models developed by enterprises make it to production.

The study found that, on average, just 54%of AI models make it from the prototype to the complete production phase. This is only a slight increase over the widely cited 53% found in a poll by Gartner in a 2020 survey.

“The biggest surprise was the sheer number of organizations that reported having thousands of AI models deployed coupled with the fact that only 54% make it into production, and many [indicating] they have not aligned to business value,” Frances Karamouzis, the distinguished VP analyst at Gartner, said in an interview.

In such a situation, what must be done to move the needle to push more AI initiatives from pilot to full-scale implementation? Karamouzis claims that the solution may be summed up in a single word: Discipline.

She believes companies need a deliberate plan to align value, recruit the right people, and implement key components of AI trust and security.

Governance is still a problem

Another result from the Gartner survey is that 40% of companies have adopted thousands of AI models. The sheer volume leads to complexities in governance and monitoring of AI’s value and Return on Investment (ROI).

Other surveys released this year have already highlighted the difficulty of AI’s lack of governance. A June 15 report produced by Juniper Networks and Wakefield Research highlighted the lack of maturity in AI governance legislation as a barrier to future adoption. However, the Wakefield Research study found lack of talent to be an area of concern, which Gartner does not consider.

An April 2022 poll published by O’Reilly Media identified governance as a barrier to AI adoption. According to the data, 51% of businesses did not have a governance framework for AI projects.

The place where security, privacy, and AI meet

Participants in the Gartner survey did not list security as a barrier to adoption. Only three percent of respondents named security as the biggest obstacle. The top barriers were the inability to assess value, a lack of understanding of AI’s benefits and uses, and data access challenges.

Even though security was not ranked as one of the biggest obstacles, there are still significant issues with AI-related security and privacy. About 41% of businesses acknowledged previously experiencing security issues at some point in the past.

More investigation into AI security found that 50% of the businesses were worried about threats from rivals and even alliance partners. However, it appears that the real danger comes from within the organization itself. Among those companies that admitted facing AI-related privacy or security issues, 60% percent admitted that they were triggered by people working within the organization.

“Organizations’ AI security concerns are often misplaced, given that most AI breaches are caused by insiders,” Erick Brethenoux, distinguished VP analyst at Gartner, wrote in a release. He added, “While attack detection and prevention are important, AI security efforts should equally focus on minimizing human risk.”

From October 2021 to December 2021, researchers from Gartner polled people in the United States, Germany, and the United Kingdom. Six hundred and ninety-nine people filled the survey; all of them work for companies planning to use AI within the next three years or that have already done so.