Dive Brief:
- Just 14% of global enterprises that deploy AI said they have a clear strategy with defined goals and outcomes for the technology, according to a report from business services company Altimetrik and HFS Research today that surveyed more than 500 tech leaders. The majority, 71%, have an incomplete or developing strategy.
- AI accountability lies most often with CIOs, CTOs or other tech leaders, and the report found that they feel pressure to deploy AI before they have systems in place to govern them, training for their employees or a plan for who is accountable when something goes wrong.
- Tech leaders cited lower operating costs as their primary driver for AI adoption, but it’s not the correct approach, said Mark Baker, a chief AI practitioner at Altimetrik. “Cost cutting is an outcome, not a strategy,” he said. “How do you find ROI? The same way as always: by figuring out what a problem is, developing a solution and understanding the savings in that.”
Dive Insight:
Industry pressure to deploy AI is leading a majority of enterprises to experiment without clear human oversight, purpose or accountability.
“I’m calling it ‘the AI moment,’ this idea that you have to do something now,” Baker said. “People are saying ‘We know the solution, it’s generative AI. Now go find the problem.’”
Generative AI does present something new to the enterprise, Baker said.
Until a few years ago, enterprise technology was built on deterministic systems, algorithms with predefined rules that did not have autonomy to work outside of the processes they were built for. Enterprises developed an understanding for who to hold accountable when these systems worked incorrectly, and they must now adapt those ideas for the probabilistic systems that make up generative AI models.
To do that in a meaningful way, enterprise tech leaders need more involvement in the building, planning and testing for those who want to deploy it, Baker said.
Most companies experimenting with AI are finding ROI, according to KPMG, but those with mature AI deployments — that have invested in their people, behaviors and trust — are reaping higher rewards.
“People involved with the conception of the project actually need to understand what we’re asking [AI] to do,” Baker said. “Once you have that, accountability and ownership will develop. It will become a part of IT’s standard operating procedures.”
Education across an organization plays a large role in successful AI experimentation, Baker said. But most enterprises don’t invest in it — nearly 80% of the survey’s respondents said their employees receive fewer than 10 hours of AI training per year. The gap is leading to a lack of confidence, the report found, with 43% of respondents saying they have self-doubt or feel behind when asked to use AI.
Just over half of respondents said they expect roles at their organization to shift or be reduced because of AI in the next few years, but most of that group expect it will happen through attrition instead of an active strategy.
Baker said CIOs and other tech leaders can take more ownership over their AI strategies, but they need to understand the specific problems they’re aiming to solve before jumping into AI pilots. Tech governance has always come from asking the right questions of the technology and checking that it meets your goals.
A lack of understanding partnered with the pressure to adopt has led to “a bit of a panic” for CIOs, Baker said.
“It’s the AI moment and the terrifying velocity that it’s coming at us,” he said. “Take a breath, do the basic work of understanding the moment and answering the questions.”
