Corporate AI adoption continues to advance, but the workforce practices needed to support it are developing at a different pace. Organizations are putting AI into production, expanding its use across business functions, and encouraging employees to work with new tools. At the same time, formal training remains limited, many companies lack a defined AI talent strategy, and executives are still determining which skills will matter most as AI becomes part of everyday work.
Those findings were at the center of the AI Leaders Council webinar, 2026 State of Corporate AI Talent Study – Panel Preview, featuring Glenn Hopper, AI author and consultant; Nirmal Jingar, AI leader at Wayfair; and Neil Brown, Executive Director of the AI Leaders Council.
The panel reviewed findings from the 2026 State of Corporate AI Talent Study, which surveyed more than 300 executives across North America from June through August. Respondents included CIOs, CTOs, Chief AI Officers, AI directors, and other executives responsible for technology, data, operations, and workforce strategy.
AI Adoption Is Moving Forward
The study found that AI has become an active concern for nearly every organization surveyed. Only 3% reported having no AI initiatives underway, compared with 13% in the AI Leaders Council’s January research.
Forty percent of respondents said their organizations were experimenting with or piloting AI, while 35% reported limited production use cases. Another 19% had scaled AI across multiple functions. Only 3% said AI was fully embedded across the organization.
The distinction between experimenting with AI and embedding it into enterprise operations became an important part of the panel discussion. Hopper described two broad forms of adoption: bottom-up use by individual employees and top-down implementation connected to company systems, data, and controls.
He cautioned that employee adoption alone does not necessarily indicate that an organization has developed a mature AI operating model.
“I think it’s great. I think there’s a lot of efficiencies to be gained here, but we’re also in this kind of a dangerous stage with AI adoption unless companies have gone through and put these guardrails in and have the controls and all that in place.”
For larger enterprises, Jingar noted that adoption becomes considerably more complicated when AI reaches systems that affect finance, banking, healthcare, supply chains, customers, and company revenue. Governance, security, and appropriate guardrails become essential as organizations move AI into these environments.
AI Usage Varies Considerably Across Business Functions
IT and engineering currently lead organizational AI usage at 65%, according to the study. Operations followed at 49%, executive decision-making at 45%, sales and marketing at 42%, finance and accounting at 35%, and customer support at 31%.
The differences reflect more than the availability of technology. They also reflect the requirements and risks associated with each function.
Hopper pointed to finance as an example. Finance teams have good reason to proceed carefully because errors can affect reporting, controls, and audits. As AI technology improves, he expects existing finance controls to become an important foundation for broader adoption.
Jingar identified another constraint: organizational knowledge is frequently undocumented. Years of domain expertise, decisions, and business context may reside with employees rather than within systems that AI can access. Companies therefore need to consider data quality, documentation, knowledge graphs, and other methods of providing AI systems with appropriate business context.
Employee AI Usage Remains Uneven
Enterprise adoption does not necessarily mean that most employees use AI regularly.
Twenty-nine percent of respondents estimated that between 10% and 25% of employees actively use AI tools in their daily work. Another 25% said fewer than 10% of employees were regular users. Together, 54% of respondents estimated that no more than one-quarter of their employees regularly use AI. Only 6% reported AI usage among more than three-quarters of employees.
Hopper suggested that usage percentages are useful for understanding the current adoption cycle, but they may eventually become less meaningful than business results.
“Ultimately, that’s going to be the wrong metric. This is giving us insight into how it’s being adopted, but ultimately, I don’t care if you use an abacus or AI. It’s about the output that you get out of it.”
That distinction gives executives another way to evaluate AI programs. Tool usage can indicate whether employees are experimenting with the technology, but productivity, quality, accuracy, and business outcomes provide a more substantial measure of whether that adoption is producing value.
Upskilling Is the Leading AI Talent Strategy
The study also examined how organizations intend to build the skills required for expanded AI adoption.
Upskilling existing employees was the most common strategy, cited by 44% of respondents. Yet 33% reported having no defined AI talent strategy. Seventeen percent were combining external hiring with internal upskilling, while only 4% were primarily hiring external AI talent.
For Hopper, one of the fundamental questions is what organizations actually mean when they refer to AI skills. Training employees on the latest tools is useful, but companies also need to consider process knowledge, documentation, human oversight, and the division of responsibilities between employees and machines.
His recommendation was straightforward: “You can’t upskill into a capability that you haven’t defined.”
Organizations first need to understand their processes, determine which activities can appropriately be handled by technology, identify where human involvement remains necessary, and then develop training around those decisions.
Jingar described several distinct skill groups that organizations should consider. Broad AI literacy and fluency can benefit the workforce generally. Domain experts need to understand how AI applies to their particular business responsibilities. Builders require the engineering and technical skills to create production AI systems. Leaders need sufficient knowledge to determine where AI should be applied, how work should be redesigned, how risks should be managed, and how business impact should be measured.
Most Companies Are Not Hiring Dedicated AI Specialists
Despite growing adoption, 73% of organizations surveyed said they were not currently hiring for AI-specific positions.
Among organizations that were hiring, AI engineers and machine learning engineers were the most frequently cited roles at 12%. AI governance and risk positions followed at 7%, while data scientists accounted for 6%.
The results suggest that companies are relying substantially on their current workforce rather than attempting to build entirely new AI teams.
At the same time, specialized hiring remains important in certain areas. Jingar pointed specifically to AI engineers capable of building AI workflows and to governance, risk, and security expertise as organizations put more agents and agentic workflows into production.
The panel also questioned whether some roles that received considerable attention during the early generative AI period will endure as standalone positions. Prompt engineering was one example.
Hopper observed: “I think that AI skills are going to be built into whatever role that we’re doing going forward.”
That prospect shifts the talent discussion from the number of dedicated AI jobs an organization creates to the skills employees throughout the company will need to perform their existing roles differently.
The Largest AI Talent Gap Is Business Translation
One of the study’s more instructive findings concerns the capabilities organizations believe they are missing.
Business translation and use case design ranked as the largest AI talent gap at 31%. Employee adoption and training followed at 24%, while data readiness and data engineering accounted for 16%. Technical expertise in machine learning and engineering, along with AI governance and risk, each registered at 15%.
For Hopper, the prominence of business translation is understandable because responsibility for it is frequently unclear.
“Business translation and use case design is the biggest one on there for a reason. And well, one, it’s kind of the only item on the list that nobody owns.”
Organizations may have defined ownership for data, technology, and governance, but deciding which AI projects are worth pursuing can fall between established functions. Effective AI programs therefore require people who understand both what the technology can accomplish and where its application can produce worthwhile business results.
Critical Thinking and Workflow Design Take Priority
When respondents were asked which skills would be most important for the workforce in the AI era, critical thinking and validation ranked first at 65%. Automation and workflow design followed closely at 61%. AI tool usage was considerably lower at 33%, followed by data literacy and change management.
The findings indicate that workforce preparation extends well beyond learning how to operate a particular AI product.
Employees need to evaluate outputs, understand processes, solve ambiguous problems, coordinate across functions, apply domain expertise, and decide when human judgment should override or supplement automated work.
As Hopper explained, many organizations may already have people with these abilities, although they have historically described the skills differently.
“The most critical skills are yes, you have to understand the technology, but where does the human in the loop kind of fit in with this?”
For executives developing AI talent strategies, the task may therefore involve identifying existing capabilities within the workforce and adapting them for an environment in which employees increasingly work alongside automated systems.
Formal AI Training Has Considerable Room to Grow
Despite increasing AI adoption, structured training remains limited.
Forty-two percent of respondents described their AI training as informal or ad hoc. Twenty-one percent offered no training, 20% provided training for selected roles, and only 17% offered company-wide training.
Employees are filling some of that gap themselves. Fifty-six percent of respondents said employees primarily learn AI through self-directed efforts. Internal training programs accounted for 23%, peer training for 10%, external certifications for 7%, and vendor training for 4%.
Both panelists emphasized the difficulty of developing training around tools that change quickly. Tool instruction has value, but organizations also need durable training around governance, security, responsible usage, and the practical application of AI within specific business processes.
AI Is Changing Jobs More Often Than Eliminating Them
The study provides a measured view of AI’s current workforce impact.
Fifty-one percent of respondents reported no significant workforce impact yet, while 37% said AI was changing existing roles. Only 6% reported that AI was creating new positions, and another 6% said it was reducing headcount.
For the panel, these findings were consistent with what they are seeing in practice. AI can already perform particular tasks well, but that is different from replacing an entire position with all of its responsibilities, context, judgment, and interpersonal requirements.
Jingar emphasized that human judgment remains particularly important because employees possess accumulated context that current AI systems often lack.
“We have so much of context in our head and we can make the judgment on the fly sitting in a meeting room, right? Or discussing with somebody, which agent is not able to do.”
The more immediate workforce question may therefore concern how jobs will change as particular tasks become automated and employees assume different responsibilities.
Governance Becomes More Important as AI Gains Authority
The panel concluded with a discussion of the risks associated with broader AI deployment, particularly as organizations move from conversational tools toward agents capable of taking actions.
Jingar stressed the importance of governance, policies, security, roles, and guardrails when AI systems are operating in production.
“Use the AI intelligence, but don’t leave the agent’s authority unchecked.”
He also emphasized the need to trace the actions an agent takes in production. As organizations grant AI systems more authority to execute work, visibility into those actions becomes increasingly important.
Hopper similarly raised the question of who owns AI governance within an organization. Policies, procedures, controls, and clear accountability will become more consequential as AI moves deeper into business processes.
Preparing the Workforce for the Next Stage of AI Adoption
The 2026 State of Corporate AI Talent Study suggests that companies have moved well beyond deciding whether they will use AI. The more consequential questions now concern how organizations will develop employees, select appropriate use cases, govern increasingly capable systems, and measure whether AI is producing worthwhile results.
The research also suggests that AI workforce planning should not be confined to recruiting technical specialists. Organizations will need technical expertise, but they will also need employees who understand their businesses well enough to translate AI capabilities into practical applications. Critical thinking, validation, workflow design, domain expertise, and judgment remain central to that work.
As adoption progresses, the companies that develop these capabilities deliberately will be better prepared to move from experimentation toward sustained operational use.
Watch the full 2026 State of Corporate AI Talent Study – Panel Preview webinar to hear the complete discussion and additional observations from Glenn Hopper, Nirmal Jingar, and Neil Brown.
You can also download the full 2026 State of Corporate AI Talent Study for the complete research findings.