When ChatGPT launched in late 2022, I instinctively knew that the era of traditional knowledge work would soon be over. It was only a matter of time. Rather than resist the inevitable change, I chose to lean into it and pivoted into AI education and consulting in spring 2023. I started teaching young students on Outschool how to master ChatGPT and other generative artificial intelligence (AI) tools like Midjourney before expanding my training and consulting to small and mid-sized businesses.
However, I wasn’t ready for how quickly and unevenly AI’s impact would reshape the business landscape. Within months, “AI” had become the trendiest buzzword, with organizations of all sizes scrambling to figure out their AI strategy based on the visions of soaring productivity, lower costs and sky-high profits sold to senior executives in boardrooms across America and around the world by consulting mega-corporations. But today, some businesses investing in their AI transformation have yet to achieve their desired goals despite the large sums of money spent.
As Sara Davison, the co-founder of AI BuildLab, observes, “A lot of the hype around the race to adopt the technology was probably going a lot faster than the maturity of the people understanding how to leverage this technology the best.” This is the AI productivity paradox at work. Massive hype and massive investment chasing a trend that yields minimal tangible ROI for some companies. A 2024 research report by The Upwork Research Institute found that 77% of employees say AI tools have decreased their productivity and increased their workload.
In this article, I’ll share how organizations can overcome this paradox. We’ll explore practical, human-centered strategies like human-AI collaboration skills, targeted middle-management engagement and workforce development that produce measurable outcomes rather than simply adopting the latest flashy technologies.
Focus on human-AI collaboration skills
One of the most effective approaches to implementing AI in an organization is redesigning workflows around the complementary strengths of humans and AI rather than attempting wholesale replacement of workers. There’s the misconception that AI can replace entire job functions overnight. That may be possible in the not-too-distant future, but powerful tools like ChatGPT and other popular consumer and enterprise AI need human collaborators and architects working behind the scenes to direct, refine and verify their output to ensure consistent quality over time.
“One of the big problems is not clearly defining what implementing at scale means. What does success look like when you unlock a bunch of licenses for your team? If you buy a thousand licenses for your organization, that’s not implementing AI. That’s purchasing a tool…. ” says Josh Huston, AI consultant and founder of Quick AI Wins. “What most people need is to learn how tools like ChatGPT and Copilot can help them in their day-to-day, which doesn’t include programming anything. It includes them opening the app, knowing what the features in that app can do for them, and knowing what they can ask that app to do.”
An excellent example of human-AI collaboration in action is the University of California, San Francisco (UCSF) Mirai AI System. According to UCSF, the AI helps radiologists provide personalized risk assessments by analyzing mammograms to predict breast cancer risk in seconds. The system can detect subtle tissue patterns that may not be discernible to the human eye. Researchers at UCSF report that this enables timely interventions like additional screenings for high-risk patients, potentially detecting cancers that might be missed by traditional screening guidelines. Redesigning a doctor’s workflow to include this technology and others like it leverages the strengths of both humans and machines rather than simply adding AI to existing processes.
Leverage middle management as AI champions
One of the biggest reasons companies fail to achieve significant productivity gains from AI is that many senior leaders do not fully understand its capabilities and limitations. Rather than a simple divide where AI excels only at routine tasks and struggles with all complex work, the reality is more nuanced. The most successful implementations match AI to the right types of complexity. For example, AI code editors like Cursor, in collaboration with Claude, can migrate entire codebases from one framework to another in hours instead of weeks, even handling complex technical implementations. However, these same systems may struggle when faced with novel problems requiring creative problem-solving or dealing with ambiguous requirements where human judgment and contextual understanding of a business are critical.
When asked about the common obstacles that affect the full-scale implementation of AI in an organization, Tyler Fisk, also a co-founder of AI BuildLab, notes that “it comes back to training and usage because ultimately, we can build out a really good system, but if the people who are going to be using the system aren’t necessarily well trained or even if they’ve gone through the training and it’s not really resonating or it’s not landing with them, that’s going to cause problems.”
This underscores the vital role of middle management in successful AI adoption. Middle managers who embrace AI rather than see it as a threat are valuable to their organizations because they can identify valuable implementation opportunities while mitigating risks. They serve as a bridge between technical teams and leadership, helping senior leaders achieve their productivity goals through targeted AI applications.
However, the rise in the use of AI in business presents a double-edged sword. While those who become AI champions will thrive, recent research by Gartner suggests many organizations will use AI to “flatten their organizational structure” and “eliminate middle management” positions. This rapid evolution is typical in AI, where major milestones now occur within weeks and months, making long-term predictions challenging. Nevertheless, the trend shows that middle managers can remain relevant in increasingly streamlined companies if they evolve into AI-fluent strategic leaders rather than mere coordinators.
Develop an AI-fluent workforce through practical application
The final key to overcoming the AI productivity paradox is developing an AI-fluent workforce. In my experience, traditional approaches to AI training often focus on technical concepts that overwhelm non-technical employees, or generic workshops that don’t address specific job functions. This aligns with the Upwork report’s findings that “47% of employees using AI say they have no idea how to achieve the productivity gains their employers expect.” A much better approach is for companies to design AI talent development programs for specific roles and functions so their employees can get practical hands-on training on how to use tools like ChatGPT, Claude and other generative AI tools to do their jobs better, reduce errors and avoid burnout.
This applied learning approach, when guided by AI specialists who understand both the technology and the specific business context, produces employees who understand AI’s capabilities and limitations within their particular roles. As a result, the workforce grows to see AI as an influential collaborator rather than a threat or magic solution to every problem. When done right, this shift in mindset will help organizations achieve the desired productivity gains.
Photo from Stokkete/Shutterstock.com
