From Tools to Teammates: Navigating the New Human-AI Relationship

By Gabby Burlacu and Kelly Monahan

Executive summary

AI is finally delivering on its long-promised productivity gains—but not without consequences. As organizations rapidly adopt AI tools, many workers report improved productivity, but also rising burnout, emotional disconnection, and a shift in how they relate to both teammates and technology. In the age of AI, sustainable productivity gains require not just AI adoption, but a redesign of how work is structured, human relationships are supported, and talent is deployed.

Key takeaways:

• AI is delivering real productivity gains—but it’s also creating new risks. Full time employees getting the most done with AI are also the most burned out, disengaged, and disconnected from their teams.
  
  • 77% of executives report gains from AI adoption, and employees say they’re 40% more productive using AI tools. But the most productive AI users are also 88% more likely to be burned out, disengaged, and twice as likely to quit.‍
    
• Legacy work systems weren’t built for Human+AI collaboration. AI is now a teammate, not just a tool, and workers are starting to bond with AI more than with each other. Yet, most organizations are optimizing for output over connection, compounding burnout and talent loss, and risking performance over the long-run.
  
  • 90% of workers see AI as a coworker. Among top AI performers, 67% trust AI more than colleagues, and 64% say they have a better relationship with AI than with their teammates.
• ‍Freelancers are modeling a more sustainable human + AI future. Freelancers are just as productive with AI, but more in control, more resilient, and more focused on leveraging AI for their own personal growth and development.
  
  • 88% of freelancers say AI has positively impacted their careers, enabling greater specialization and upskilling.‍
    
• The opportunity: Redesign work for the age of AI. As AI becomes a “coworker,” organizations need to think beyond simply deploying AI, and instead reimagine work for human-centered, AI-empowered talent, teams, and workflows that unlock sustainable performance. 

Calls to action for business leaders: 

To ensure that AI fuels long-term gains, leaders must:

1. Design work for humans + AI: Prioritize autonomy, trust, and psychological safety to foster and maintain genuine human connection as workers engage with increasingly human-like AI tools.
2. Build flexible talent ecosystems: Combine full-time employees, freelancers, and AI capabilities to create agile, resilient, and high-performing teams.
3. Redefine AI strategies: Shift from tool deployment to cross-functional approaches focused on the end-to-end experience people have with AI, complete with new roles, norms, and governance models that reflect the new dynamics of human-AI collaboration.
   ‍

AI’s productivity payoff comes with a cost

Good news—AI is finally delivering on its promise, with workers reporting a 40% increase in productivity from using it. But, these productivity gains come with a cost. 

Every morning, Adam, a director of marketing at a large agency, opens his laptop not to Slack or email, but to ChatGPT. “It doesn’t judge. It just helps,” he says, half-smiling. After toiling away alone on another late-night deliverable with no teammates available to check in with, he realizes: AI is his most consistent teammate—perhaps his only one.

While business leaders may celebrate this long-awaited productivity boost, workers like Adam reveal the other side of the story. The top AI performers, those most skilled in harnessing AI tools, are exhausted. They’re fraying at the edges of their networks—not only from work demands, but also from a growing sense of emotional disconnection.

New research from The Upwork Research Institute found that workers who are the most productive with AI are also the most burned out and disconnected from organizations and their colleagues (see Figure 1). Fully 88% of these workers report burnout, and are twice as likely to quit. Eighty-five percent are more polite to AI than humans. Sixty-seven percent trust AI more than their coworkers. And 64% have a better relationship with AI than their coworkers.

This isn’t just about tech adoption. It’s a social signal.

Figure 1.

‍

This troubling trend from our recent survey of 2,500 executives, employees, and freelance talent signals a hidden, serious impact from what is otherwise an AI success story. 

While 77% of C-level executives report AI productivity gains in their organizations, they’re facing the systemic deterioration of employee connections. These long-term costs can lead to pervasive employee disengagement and critical talent loss, which can be detrimental in a growing skills shortage– and at a time when our research shows that leaders trust the work output of humans more than that of AI. 

This productivity paradox may be a natural growing pain of today’s traditional work system, which values and incentivizes increased output with AI but fails to account for the relationships among people using it every day. With the growing presence of AI agents in the workforce, companies need to reframe how AI is integrated with their teams to foster AI-empowered talent, not burned out talent. AI isn’t simply a means of gaining efficiencies, but is rewiring the way we work together. And this rewiring needs to be accounted for:

• Over three-quarters of executives, full-time workers, and freelancers agree AI agents will completely reinvent the way people work. 
• Nearly one in three (31%) executives report their company has moved beyond exploring and piloting agent use cases to actually implementing agents in business operations. The top reported use cases for AI agents include streamlining data, customer service, and administrative processes. 
• And the hunt for talent proficient in developing and operating alongside AI agents has already begun: monthly searches for Upwork talent skilled in AI agents have grown nearly 300% over the past six months, as of May 2025. This mirrors the 85% of Upwork clients hiring for writers, translators, or developers who say they are interested in integrating AI agents. ¹ 

While AI tools are already prevalent in 86% of organizations, it is clear that AI agents are on their way to becoming an integral part of every workforce. In this report, we discuss how human-AI relationships are evolving in today’s workplace and what this tells us about the unintended consequences of our current approaches to AI. We then share how leaders can plan for a more sustainable productive future by considering not just humans + AI, but humans + AI + redesign—a holistic approach for building teams, fostering trust, and designing for productive human-AI dynamics to drive sustainable growth in the age of AI.

The productivity paradox: connection erodes as output increases

The employee-reported 40% increase in productivity due to AI tools marks a notable improvement from a year ago, when Upwork research showed AI adding to the workloads of over three-quarters of employees. So on the surface, workers are finding the right integration points with AI and applying them to everyday work. 

How did they get there? Employees say:

• They’ve had more time to experiment (30%)
• The tools themselves have improved (25%)
• They’re taking initiative to upskill themselves (22%)
• Organizations are providing the right training and incentives (22%)

Plus, employees who have seen the greatest productivity gains with AI are receiving all the right training and tools that they need from their employers (48% vs. 22% of others). 

Providing room to test and learn, along with ample training, clearly improves productivity. But those don’t address deeper, less obvious dimensions of the work experience. 

That 88% of the most productive AI-enabled workers report burnout is a red flag leaders cannot ignore. 

Employees who excel with AI tools are twice as likely to consider leaving their organizations, pointing directly to the pressure of their overwhelming workload. 

The research also examined whether employees and employers have a shared understanding of their company’s AI goals. Ironically, the most productive AI workers are the most disconnected from their organization’s AI strategy (see Figure 2). 

Figure 2.

‍

To unpack this paradoxical finding, we interviewed 16 corporate leaders to understand why high AI productivity doesn’t translate into connection to the overall company AI strategy. 

We learned that highly engaged users—even leaders—may see AI as a strategic necessity, yet feel their organization isn’t implementing it with the right rigor, caution, or understanding.

This dynamic was clearly illustrated in the contrast between Holland and Adam, two business leaders we interviewed. While Holland, a company co-owner, struggled with a company that barely touched AI—saying his partner’s “archaic approach to certain things has probably cost us”—others, like marketing director Adam, found themselves in organizations swinging too far in the other direction. 

Adam described his company’s owner as someone who “thinks every new AI function will propel us to the next level with minimal effort,” a mindset with which he strongly disagrees. “It’s going to help, and it can help. But it’s not the end-all be-all,” he said. For Adam, the challenge wasn’t resistance to AI but rather overenthusiasm without a grounded understanding of what AI can and can’t do. 

These examples underscore a core tension. Power users often find themselves stuck between two extremes: leaders who don’t get AI at all, or leaders who believe AI will magically solve everything, leaving those with the most hands-on experience struggling to navigate strategic implementation in either case.

Indeed, our research shows that middle managers are more likely than individual contributors to achieve greater personal productivity with AI, with average gains of 42% compared with 36%. 

But, this same group—caught between their leaders’ objectives and their own AI use, while also having to manage how their teams integrate these tools in alignment with corporate policies—is also more likely to experience the accompanying burnout (76% vs. 69%) and preference for interacting with AI over people (50% vs. 39%), pointing to their increased isolation amid rising disengagement.  

Workers are developing social relationships with AI

What’s most shocking are findings about workers not only turning away from the business, but also from their (human) coworkers. Instead, they’re looking to AI teammates for connection.

Eighty-five percent of those most productive with AI report they’re more polite to AI than the people around them, and 79% say AI is more polite to them. 

And as Figure 3 shows, two-thirds (67%) say they trust AI more than the people they work with, 64% report a better relationship with AI than with their human coworkers, and more than half (54%) say AI is more empathetic.

Figure 3.

‍

Even as 90% of workers now view AI as a coworker, our client research shows that genuine human connection and trust remain essential to fostering engagement, well-being, and sustainable performance– the human performance leaders continue to require in fields as wide-ranging as project management, product design, graphic and presentation design, corporate and contract law, and finance. These findings illustrate that workers achieving the greatest productivity with AI have lost a sense of psychological safety and team connection that is foundational to their work experience, fueling their burnout and intentions to leave their current employment.   

But they aren’t the only ones developing a social relationship with AI. It turns out that all workers, not just the most productive with AI, report human-like relationships with these tools. Nearly half of them say “please” and “thank you” with every request, and 87% are phrasing requests as if to a human coworker, at least some of the time.

These relationships and dialogue patterns exemplify a level of anthropomorphism—a personification of technology—that is largely absent in the AI-productivity discussion. So when organizations design for greater productivity with AI by reimagining workflows and measuring productivity outcomes, they often miss the nuanced and peculiar relationships people are clearly developing with AI (see Figure 4), and how those relationships could be compensating for fraying connections between people.

Figure 4.

‍

Several participants in our interviews expressed a preference for interacting with AI over colleagues in specific workplace scenarios, particularly when it came to accountability and idea generation. This wasn’t due to a lack of interpersonal skills but rather the emotional neutrality AI offers. 

As Stephen, a general director of a professional services organization, put it: “It’s more neutral when it’s coming from an AI than coming from a person,” referring to how AI-generated task reminders felt less confrontational than direct messages from teammates. 

Others noted that AI provided a safer space for ideation. “Sometimes I’m stuck, and I don’t want to bother someone else. AI just gives me something to work with,” explained Kate, senior leader of a life sciences organization, highlighting the low-pressure dynamic AI creates. 

In these cases, AI wasn’t just a tool—it was a frictionless collaborator, trusted precisely because it was nonjudgmental and emotionally disengaged. This subtle dynamic helps explain why some users may appear more polite or trusting toward AI than toward human colleagues. But it also points to a need to design not only for new technology and workflows, but also for the broader, interpersonal work experience.

This shift in workplace dynamics raises a critical question: What happens when the emotional comfort of AI begins to outweigh the value of human connection? To understand what a healthier, more empowering relationship with AI might look like, we can turn to a different group: freelancers.
‍

Freelancers offer a blueprint for thriving with AI

Interestingly, our research shows that freelancers are gaining as much productivity with AI as full-time employees do. But more of them—88%—say AI has positively impacted their career. There are some key nuances in how freelancers relate to AI tools and broadly experience work that explain this difference.

Figure 5. 

‍‍

‍

First, previous research shows that freelancers performing skilled knowledge work use AI more, and are more proficient in embedding AI tools into their workflows compared to their full-time employee (FTE) counterparts. But while FTEs are developing a broad range of social relationships with AI, freelancers primarily use AI as a learning partner. Ninety percent say a major advantage of using AI is that it helps them learn faster, and 42% credit AI with helping them specialize in their work. And this is preparing them for what’s next: 68% of Upwork freelancers that deliver services like writing, translation, and mobile and web design say they are interested in taking on the challenge of working alongside AI agents.¹ 

‍

Second, freelancers feel in greater control of work outcomes, with 34% strongly agreeing that AI gives them an extra edge in their work versus just 28% of FTEs. And it’s freelancers’ sense of control—or self-efficacy, the belief in their own capacity to accomplish work and influence outcomes—that’s driving a connection between their AI productivity and career growth. 

Full-time employees generally do not experience the same sense of control. Forty-eight percent say their employer is asking too much of them when it comes to integrating AI. And although 30% of FTEs say AI has helped them take on new projects, considerably fewer credit AI for helping them achieve faster promotions, better pay, or better job opportunities. 

Freelance talent is key to the discussion of how work can be redesigned for a better, more productive human + AI future. Redesigning talent models by blending in-house workers and freelance talent opens access to people who are highly proficient in AI. It also enhances teams through embedded proof points and examples of effective human-AI relationships that focus on human learning rather than social replacement. What we learn here is that encouraging AI-literate talent is not enough; it's AI-empowered talent that will be the key ingredient in this work redesign.
‍

Human + AI + redesign: the formula for sustainable productivity

As AI becomes ever-present in our work, on our teams, and in our lives, it’s time to refactor the entire work experience. Taking on a workforce transformation requires confident and curious leadership to channel three workstreams for impact:

1. Design work for humans + AI: Prioritize autonomy, trust, and psychological safety to foster and maintain genuine human connection as workers engage with increasingly human-like AI tools.
2. Build flexible talent ecosystems: Combine full-time employees, freelancers, and AI capabilities to create agile, resilient, and high-performing teams.
3. Redefine AI strategies: Shift from tool deployment to cross-functional approaches focused on the end-to-end experience people have with AI, complete with new roles, norms, and governance models that reflect the new dynamics of human-AI collaboration.

Redesigning the workplace for an AI-integrated future isn’t a thought experiment—it’s a strategic imperative. Hiring managers on Upwork generally agree that an overwhelming majority of work will continue to be human-led even five years from now, with 47% of all work output projected to come from human-AI partnerships and an additional 27% from humans alone.² Research consistently shows that employees thrive when they have autonomy, psychological safety, and growth opportunities. Yet these are precisely the areas under strain in high-AI environments. 

According to self-determination theory, people are most motivated when they feel competent and in control. A new framework is needed—one that begins with agency. 

Equally important is developing a new workforce model. Today’s most agile teams are made of more than full-time staff alone. They’re blended teams, integrating freelance specialists, external partners like managed service providers, and AI systems in dynamic configurations capable of executing complex projects and driving growth. Freelancers already show how AI can be a catalyst for specialization, learning, and empowerment. Organizations should look to this model as a future-forward blueprint.

Finally, strategy must reflect the reality that AI is not emotionally neutral for workers– in other words, it’s not just a tool. Anthropomorphism isn’t a quirk—rather, our findings suggest it could be a coping mechanism in environments lacking trust. Leaders must invest in relationship-centered design, ensuring that people feel seen, safe, and supported even as digital coworkers grow more prevalent. 

The tools, tech and even who and what we define as a teammate are constantly evolving and expanding in the age of AI. But the heart of work remains connection. Productivity gains are only sustainable when AI augments—not replaces—human connection, purpose, and growth.The organizations that thrive will be those that do more than simply measure productivity; they’ll design for optimal, sustainable relationships—both human-to-AI and human-to-human.

‍

Methodology 

Research findings are based on a survey conducted by Walr, on behalf of Upwork and Workplace Intelligence, between March 24 and April 9, 2025. The survey targeted respondents in the U.S., U.K., Australia, and Canada. In total, 2,500 global workers completed the survey, including 1,250 C-suite executives; 625 full-time, salaried employees; and 625 freelancers. All respondents were between the ages of 18 – 78, were required to have at least a high school diploma, and were required to use a laptop or computer for their work at least “sometimes.” Employees in the top quartile for self-reported AI-related productivity gains were compared to their peers along dimensions of burnout, intentions to quit, and preference for AI tools vs. human colleagues. 

Additionally, focus group interviews were conducted in April and May of 2025 with 16 U.S. business leaders representing a range of industries, company sizes, and varying stages of AI adoption, and 24 U.S.-based freelance workers using AI within their work at least several times per month. 
‍

About the authors 

Gabby Burlacu

Dr. Gabby Burlacu is Senior Research Manager at Upwork, where she studies how organizations are adjusting their cultures and talent practices to access skilled talent in a rapidly evolving world of work. Her research has been featured in a variety of peer-reviewed studies, articles, book chapters, and media outlets, and has informed strategy and technology development across a range of Fortune 500 companies. Gabby holds a Ph. D. in industrial-organizational psychology from Portland State University.

Kelly Monahan

Dr. Kelly Monahan is the Founder and Managing Director of the Upwork Research Institute, where she leads research on emerging technologies, remote workforce strategies, and fostering inclusive cultures for non-traditional talent like freelancers. With over a decade of experience in future of work research, her work focuses on delivering actionable insights to help organizations adapt to the evolving world of work.

Kelly is the author of two books, including the USA Today bestseller Essential, and How Behavioral Economics Influences Management Decision-Making: A New Paradigm. She holds a B.S. from Rochester Institute of Technology, an M.S. from Roberts Wesleyan College, and a Ph.D. in organizational leadership from Regent University.
‍

About The Upwork Research Institute 

The world of work is not the same as it was just a few years ago and leaders are facing brand new challenges as a result. The old work playbook is gone, and in its place, there are debates and decisions around workforce location, worker arrangements, and flexibility. However, leaders do not need to navigate this new world of work on their own.

The Upwork Research Institute is committed to studying the fundamental shifts in the workforce and providing business leaders with the tools and insights they need to navigate the here and now while preparing their organization for the future. Using our proprietary platform data, global survey research, partnerships, and academic collaborations, we will produce evidence-based insights to create the blueprint for the new way of work.
‍

Acknowledgments

The Upwork Research Institute would like to thank Anna Brown, Dan Schwabel, Rebecca Scott, and Christine (Kim) Lee for their contributions to this research report.

‍

‍

¹ Source: AI agent awareness and interest survey conducted in Q2 2025 with Upwork clients and freelancers, primarily those hiring and working in the writing, translation, and web and mobile design work categories 

² Source: Monthly AI survey conducted with Upwork global freelancers and clients. These values represent client projections for work output five years from now, as of June 2025