In general, respondents rated their work quality higher when working with human colleagues. Respondents across all age groups reported receiving higher-quality feedback from human colleagues compared to AI. Aside from the youngest workers, most preferred peer collaboration over AI for core work tasks. These findings suggest that while workers recognize the value of AI tools, they still see benefits to human collaboration—especially for work requiring in-depth knowledge.
However, when it comes to speed of delivery, the picture shifts. Across several tasks—especially problem solving, creativity, and training—respondents from younger age groups (ages 18 to 44) were more likely to say that AI helps them deliver faster. The youngest workers (ages 18 to 24) rated AI higher on speed in nearly every task type. These findings seem to reinforce the perception that AI tools excel in efficiency, even if they don’t yet match humans in quality or nuance.
This reflects a key tension at the heart of AI integration: the tradeoff between speed and quality. While AI is valued for its ability to deliver faster results, workers across generations remain cautious about using it for work requiring human-centric skills. Monitoring how these preferences change over time will be key as technology evolves, particularly as AI-native workers entering the workforce may have stronger preferences for using these tools compared to those who entered the workforce before the advent of AI.25 Investing in AI technologies may not only enhance current workforce productivity, but it could also attract talent eager to benefit from the latest advances.26
Our survey findings also suggest that there is scope to better train and develop AI tools and agents to help workers with core tasks, creativity, brainstorming, training, and seeking feedback or advice. Respondents’ preferences for working with AI were also influenced by several other factors, including their preparedness and perception of AI’s utility. Workers who felt more prepared to use AI at work were more likely to say that AI outperforms humans in speed. Similarly, those who found AI more helpful in their work were more inclined to rate AI higher in both quality and speed. These findings emphasize the importance of investing in AI literacy and skill development programs as part of a holistic approach to technology and talent development.27
Key takeaways
Leaders can consider the following actions to prioritize human oversight of agentic AI as it becomes more embedded in organizational workflows:
Establishing clear boundaries for automation around tradeoffs between efficiency and quality. Updating AI usage guidelines to distinguish between activities where AI can be used for speed and efficiency, and those that require deliberate human oversight or governance, can help create these boundaries.Monitoring workforce preferences around human and AI collaboration. Surveys and feedback can be useful tools for tracking worker preferences. Surveyed workers rate humans higher on quality, especially for core tasks, but younger surveyed workers (ages 18 to 24) say AI delivers faster results in nearly every area. Defining where and how to use AI for work could become critical for organizations in the future, especially as digital natives majorly comprise the workforce.Keeping humans “on the loop” instead of just “in the loop.” As agentic AI takes on more complex tasks at work, organizations could shift from direct control to supervisory oversight on these tasks. Treating AI like a junior team member—monitored, guided, and corrected when needed—can help enable efficiency without sacrificing accountability.Investing in AI literacy and skill development. Respondents who feel more prepared to use AI and understand its capabilities and limitations rate it higher in both speed and quality. Prioritizing training programs that build AI fluency can help increase adoption and impact.
Opportunity 3: AI can be an upskilling engine—especially for entry-level workers
Organizations are increasingly linking together multiple agentic systems, taking agents designed to handle specific tasks and combining them to create larger workflows. These new workflows can help boost productivity across job levels and free up time for workers to focus on other tasks.
However, while agentic technology may offer efficiencies, it also raises questions around skill development, especially for early-career workers. It will likely be important for organizations to invest not only in technology but also in developing a robust talent pipeline.
Deloitte’s 2025 Human Capital Trends found that two-thirds of surveyed hiring managers and executives believe entry-level hires are underprepared, with lack of experience being the primary weakness. One risk of treating AI development separately from human capital development is that automation may reduce opportunities for on-the-job training. Automating tasks traditionally performed by entry-level workers, such as taking and distributing meeting notes, may create short-term efficiencies but may limit on-the-job training and other tools organizations have traditionally used to upskill early-career workers.28 Deloitte research suggests that the rise of AI could be contributing to the difficulty workers face in finding entry-level jobs.29
To manage this dynamic, organizations can leverage AI to accelerate skill development for early-career workers. Sixty-one percent of respondents to our survey believe that AI can support upskilling opportunities for entry-level workers, with respondents between the ages of 25 and 34 years being the most likely (66%) to agree. Research by the National Bureau of Economic Research shows that gen AI can be particularly beneficial for lower-skilled workers, offering real-time support and coaching to help them learn faster.30 It also suggests that early-career workers are more likely to embrace AI as a tool for skill development.