AI is changing how teams work together, but it’s also creating challenges in accountability. Here’s what you need to know:
AI can boost productivity and idea generation, but without clear accountability, it can lead to costly mistakes. Organizations must prioritize governance to balance AI’s potential with responsible use.
AI Impact on Cross-Functional Teams: Key Statistics and Accountability Gaps
Recent findings show that AI is reshaping how cross-functional teams collaborate. A field experiment conducted in March 2025 at Procter & Gamble with 776 professionals highlighted how AI breaks down traditional silos that often hinder teamwork. Typically, R&D teams focus on technical solutions, while Commercial teams lean toward market-driven approaches. However, when AI was introduced, both groups created more balanced and integrated proposals, regardless of their functional backgrounds [3][6].
The benefits don’t stop there. Teams using AI were three times more likely to generate top-10% ideas compared to individuals working without AI. They also reported a 30% improvement in efficiency and innovation compared to non-cross-functional teams [4][5]. Beyond these metrics, AI-enabled teams enhanced quality by 0.39 standard deviations and reduced task completion time by 12.7% [6].
"If you want to empower an individual to be as effective as a team, give them AI. But if you want to be in that top 10% of performers, a full human team plus AI seems like the recipe for success."
– Fabrizio Dell'Acqua, Postdoctoral Research Fellow, Harvard Business School [4]
AI plays a crucial role in bridging departmental divides by democratizing expertise and reducing "coordination overhead" - a term researchers use to describe the routine tasks that typically consume 60% of a knowledge worker's time. This shift allows teams to focus more on critical decision-making and creative problem-solving instead of mundane updates and data synthesis.
While these advances are promising, they also introduce new challenges, particularly around accountability.
Despite AI’s ability to enhance team dynamics, it raises significant accountability concerns. A striking 76% of executives now view AI as more of a coworker than a tool [8]. But when AI makes mistakes, responsibility becomes murky. Is it the fault of the deploying company, the development team, the model provider, or the end user? [7].
The Klarna case is a stark example of this issue. Between early 2024 and 2025, Klarna replaced 700 customer service agents with AI, reducing its workforce from 5,500 to 3,400 employees. However, this shift led to an accountability crisis. By 2025, CEO Sebastian Siemiatkowski admitted, "AI handles the tickets. Nobody handled the quality", prompting the company to quietly reassign workers back to support roles to reintroduce human judgment [4]. This situation highlights the urgent need for strong governance frameworks as AI becomes a core part of team operations.
Currently, only 20% of companies have mature governance systems in place to manage autonomous AI agents, even though 88% of organizations are already integrating these agents into their workflows as of 2026 [9]. This gap poses a strategic risk, as AI adoption outpaces the development of processes to assign accountability. Complicating matters further is the tension between maintaining human oversight and granting AI the autonomy needed for scalable operations. Balancing these elements remains a critical challenge for organizations moving forward.
One of the best ways to tackle accountability issues is to assign a dedicated team or leader responsible for overseeing AI behavior and its impact. This empowers them to make decisions and escalate concerns when necessary [13].
Companies like JPMorgan Chase, Mastercard, and Salesforce have already set up specialized AI governance teams. These teams bring together experts from various disciplines to manage both ethical and technical risks [13]. Engaging risk and compliance specialists early in the process is a critical step [13][15].
"Accountability must be designed in from the start - not bolted on after deployment"
– Wharton Executive Education [13]
Leaders of individual business units should take ownership of their AI use cases by defining goals, managing risks, and tracking outcomes. At the same time, data science teams focus on validating the technical aspects [11]. A 2025 global survey found that 77% of organizations are working on AI governance initiatives, with this number climbing to nearly 90% among those already using AI [11].
Once ownership is clearly assigned, the next challenge is ensuring AI is seamlessly integrated into everyday workflows.
After establishing leadership, teams need a clear plan for embedding AI into their operations. A practical approach involves creating shared KPIs that align technical metrics (like model accuracy) with business objectives (such as audit success rates, documentation quality, or faster risk responses) [12].
Take JPMorgan Chase as an example: by using AI-powered fraud detection and fostering collaboration between legal, compliance, and data science teams, they were able to cut fraudulent activity by 15–20% [14]. A key factor in their success was the early development of governance plans through teamwork across departments [12].
"If a model is biased, is it a technical issue? A compliance issue? A legal risk? Frequently, the answer is yes (to all of them). Which is why working in silos doesn't work"
– Governance expert from Lumenova AI [12]
To make accountability a regular part of the workflow, organizations can embed governance into development processes. This includes using automated checkpoints and audit logs during each AI sprint [12]. These structured experiments pave the way for smoother collaboration across teams.
For broader organizational coordination, many companies are forming AI councils. These councils bring together executive leaders to shape strategies, while operational teams focus on execution [16][11]. Nearly half of all organizations now rank AI governance among their top five strategic priorities [16][11].
These councils often use a "three lines of defense" model:
To avoid spreading accountability too thin, councils should assign a single leader to oversee specific outcomes [17].
"You must have a clear leadership commitment to build an organizational culture around AI governance... Responsible AI culture isn't defined by having a governance body. You create culture from the top"
– Advisory Board Member Deepti Kunupudi [11]
These councils simplify decision-making while reinforcing accountability, a recurring challenge in AI governance. This approach aligns with the concept of "systemic codependency", which emphasizes the need for interconnected leadership systems. By redesigning roles and workflows for collaborative, outcome-focused efforts, organizations can better manage the complexities of AI development [10].
In 2025, AI adoption among workers grew by 13%, but confidence in its use dropped by 18%. Even more concerning, 40% of tasks were completed with unreviewed AI outputs, signaling a growing dependence on AI tools without adequate oversight[18].
To address this, leaders need to shift their focus from purely technical training to enhancing human judgment. A good example comes from early 2026, when FedEx partnered with Accenture's LearnVantage platform to roll out an AI literacy program for over 400,000 employees. This initiative went beyond teaching how to use AI - it emphasized the importance of recognizing when human intervention is critical[18].
The key lies in blending technical skills with critical thinking. Teams should treat every AI-generated result as a starting point, not the final answer. Implementing practices like the "AI Plus One" rule - requiring a human review of every significant AI recommendation - can help maintain accountability and prevent costly mistakes[18].
By 2026, half of all organizations are expected to include AI-free skills assessments to ensure employees retain sharp critical thinking abilities[18]. Some companies are even holding brainstorming sessions without any AI tools. This isn’t about rejecting technology - it’s about keeping human cognitive skills in top shape. A Fast Company article summed it up well: "Would I stand by this if my name was attached to it?"[18].
Efforts like these to reskill teams are essential for building the trust needed to integrate AI effectively.
Trust in AI doesn’t come from blind reliance; it’s earned through deliberate calibration and transparency. Strengthening trust in AI not only fosters better teamwork but also reinforces accountability across the board. Leaders can schedule regular review sessions to examine specific AI-driven outcomes, focusing on where the technology worked well and where it missed critical details. This process helps teams build collective judgment and use AI more effectively[18].
However, skepticism remains high. Nearly 60% of workers believe AI worsens existing biases, and 79% lack trust in businesses to use AI responsibly[18]. On the other hand, research from Harvard Business School shows that while AI can boost individual productivity by 40%, cross-functional teams using AI are three times more likely to generate groundbreaking ideas than individuals working alone[4].
Clear policies are critical for moving forward. These should outline which tasks are appropriate for AI and which require mandatory human oversight[18]. For example, labeling documents with "AI-assisted draft, reviewed by [Name]" ensures transparency and accountability while maintaining the pace of work. The goal isn’t to slow things down but to protect the integrity of decision-making processes. By combining clear policies with human judgment, organizations can build a "Human Moat" that safeguards their competitive edge even as AI continues to evolve.
Leaders play a vital role in reinforcing this balance. Experts like Seth Mattison stress that blending strong human judgment with advanced AI technology is the key to staying competitive in an AI-driven world.
By 2026, AI has shifted from being a novelty in cross-functional teams to a core operational tool. Only a small fraction - less than 2% - of customer experience leaders remain stuck in the experimentation phase, while 77% are actively implementing AI at different levels of maturity [2]. However, a glaring issue persists: just 31% of organizations have established comprehensive governance policies, and about 20% lack any formal assignment of AI accountability [2]. This lack of oversight risks eroding competitive advantages.
To address this, organizations must assign clear ownership for every AI system. This includes designating specific individuals as data, model, and business owners. Without such accountability, AI can drift into "shadow use" - a term experts use to describe unauthorized deployments that bring financial and reputational risks. As Divya Parekh, Founder of The DP Group, explains:
"The structure is not the secret. When teams know who owns the vision, who owns delivery and how fast decisions get made, AI stops being hype" [19].
Governance should evolve from occasional oversight to continuous monitoring. This means implementing real-time tracking systems, bias-testing protocols, and "kill switches" to shut down systems that deviate from established policies. Past failures have already highlighted the dangers of neglecting proper oversight.
Strong governance works hand-in-hand with human accountability. Seth Mattison describes this as building a "Human Moat", emphasizing that while AI capabilities are advancing rapidly, systems of human accountability are lagging behind. As Mattison points out, the winners in this space won’t be those who simply deploy AI quickly but those who manage and discipline it effectively. Professor Saharsh Agarwal from ISB echoes this sentiment:
"The next era of leadership will be defined not by how well executives deploy AI, but by how well they discipline it" [1].
Accountability for AI-driven decisions often hinges on the organization's governance framework. Typically, responsibility is distributed across cross-functional teams, including legal, compliance, and data science departments. However, when these teams work in isolation, managing accountability can become challenging. Establishing clear roles and fostering collaboration among these groups is essential to ensure decisions are responsibly handled.
AI governance is about setting up frameworks, policies, and processes to guide the responsible, ethical, and effective use of AI. This means defining key performance indicators (KPIs) to measure governance, keeping track of return on investment (ROI), and addressing risks such as privacy concerns and fairness issues. It also involves ensuring cross-functional teams work together seamlessly, avoiding silos that can hinder progress.
Strong governance helps leaders make decisions tailored to specific contexts, encourages transparency, and strikes a balance between oversight and adaptability. The goal is to harness AI's potential while minimizing potential risks.
AI decisions that call for human review often involve ethical dilemmas, major strategic decisions, and issues tied to accountability and trust. This is particularly important in regulated industries or when AI systems operate without clear transparency or a track record of reliable performance. Incorporating human oversight in these cases helps ensure decisions are made responsibly and thoughtfully in crucial areas.