Tom Dunlop is cofounder and CEO of Summize, a CLM solution, and a former General Counsel for high-growth technology companies.
At the leadership level, I’ve seen a prevailing narrative: AI is already delivering major efficiency gains. But when you talk to the people actually doing the work, the reality often looks different.
In many cases, the work hasn’t disappeared; it has simply moved somewhere else. In fact, what I see from regularly speaking with teams and customers is that AI usage is often fairly surface-level and hasn’t fundamentally changed how work gets done.
Expectation Versus Reality: What The Data Says
The data reflects this gap between expectation and reality. Gallup research shows that workplace AI use is increasing but remains relatively limited. The share of U.S. employees using AI daily rose from 10% to 12% between 2023 and late 2025, reflecting gradual adoption rather than widespread transformation.
At the same time, research from the Federal Reserve Bank of St. Louis suggests generative AI is already producing measurable efficiency gains. Among workers who actively use generative AI, the technology saves an average of 5.4% of their working hours. However, when those gains are averaged across the entire workforce, including employees who do not yet use AI, the impact drops to 1.4% of total work hours saved.
Those numbers show AI is helping people work more efficiently, but the gains so far are incremental rather than transformational.
That’s okay.
A big misconception about AI is that it should immediately automate entire workflows. Most successful AI implementations today are relatively simple, serving as companions to people doing the job by summarizing information, drafting documents, analyzing data and accelerating repetitive tasks. Those capabilities can remove significant friction from everyday work, but AI still relies on human judgment, context and oversight.
Building Trust In AI
Trust is one of the most overlooked factors in AI adoption. Employees need confidence that AI systems produce reliable outputs. Leaders need confidence that the technology is operating on accurate information. And organizations need assurance that AI is aligned with existing workflows rather than creating new risks.
Trust is built on three elements:
1. High-Quality Data: AI systems are only as reliable as the information they draw from. When organizational data is fragmented, outdated or poorly structured, the results are often inconsistent. Before expanding AI into critical workflows, leaders need to understand whether the underlying data environment can support it.
2. Access To Institutional Knowledge: Some of the most effective uses of AI are not about automating decisions, but about making expertise easier to access. In most organizations, knowledge is scattered across conversations, emails, documents and systems, where it’s often disconnected from the work itself. Employees don’t just need help finding information; they need help understanding what matters, in context, at the moment they’re making decisions. When AI can surface and apply that knowledge in a meaningful way, it can reduce friction and enable more consistent work across the organization.
3. Gradual Deployment: Organizations that want to see the most success with AI should not start with sweeping automation. Instead, they must focus on targeted use cases where the benefits are clear. Think summarizing large documents, drafting internal content and automating routine information handoffs between teams.
These smaller applications can help employees become familiar with the technology and provide leaders with insight into where AI fits best.
Automate Tasks, Not Entire Jobs
Most jobs involve some level of analysis, communication and creative judgment, along with repetitive tasks that consume time but do not require deep expertise. AI tends to perform best by handling the latter.
Consider a common example involving a sales role. Much of a salesperson’s value comes from building relationships, understanding customer needs and navigating complex conversations. Those parts of the job rely heavily on human judgment and experience.
But sales roles also involve substantial administrative work: updating CRM systems, summarizing meetings, logging follow-ups and preparing reports. These are the tasks where AI can remove friction and give employees time back.
The same pattern appears across many roles, including legal teams. Lawyers are hired for their judgment, interpretation and advice. Yet, much of their time can be spent reviewing documents, summarizing information or navigating large volumes of text. AI can accelerate many of those tasks without replacing the core expertise required by the role.
This is where the narrative around AI needs to change: It’s not about job replacement, but task replacement. Future job descriptions will likely include expectations around how employees use AI tools to support their work, whether that means generating notes, updating records or analyzing data to surface patterns, risks or opportunities that would otherwise be difficult to identify.
AI Adoption Is A Culture Challenge
The biggest determinant of whether AI improves productivity is leadership. Leaders shape how teams approach new tools, influencing whether experimentation and adoption are encouraged or avoided. They determine whether AI becomes embedded in everyday workflows or remains a separate initiative used by only a small group of early adopters.
The organizations that want to see the most success with AI must treat adoption as a people challenge as much as a technology one. That means not only communicating clearly about how AI will be used, where it will support employees and what expectations exist around its role in daily work, but also creating an environment where teams feel comfortable testing and learning. Encouraging experimentation within clearly defined parameters can allow employees to explore how AI can support their work without introducing unnecessary risk.
Turning Incremental AI Gains Into Real Productivity
The modest productivity gains reported so far should not be seen as disappointing. They reflect the reality that organizations are still learning how to apply AI effectively. Technology rarely transforms work overnight. Progress comes through smaller improvements that compound over time.
The leaders who want to see the greatest benefits from AI must focus less on replacing roles and more on removing friction from the work those roles perform. They must introduce AI to support employees, redefine how tasks are completed and gradually embed the technology into the work culture. The organizations that recognize this early will be better positioned to turn incremental improvements into lasting productivity gains.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
