The Impact of Generative AI on Knowledge Workers and Critical Thinking

Over the past two years, I’ve seen first hand for myself and my clients the positive impact generative AI is having on knowledge workers. These professionals deal with unstructured information: data that, until recently, couldn’t be effectively summarized or processed.

Multiple studies and reports confirm that AI tools are driving productivity gains:

• A survey of Australian government departments found that 69% of users reported Microsoft’s Copilot helped them complete tasks faster.

• Research by Nielsen Norman Group analysed three case studies and found that generative AI boosted business user productivity by an average of 66%.

• Forrester found that small and medium businesses (SMBs) achieved a return on investment (ROI) between 132% and 353% with AI tools.

• The quarterly journal of economics found generative AI helped customer support agents resolve 15% more cases per hour on average.

While generative AI undeniably enhances productivity, we must consider a crucial aspect of the role that effective knowledge workers perform every day: critical thinking.

I previously wrote about the importance of original thought in the AI era. Here, I want to expand on that by exploring how AI influences critical thinking, an essential skill for knowledge workers, and what we need to consider when designing generative AI solutions.

The Role of Critical Thinking in AI-Driven Work

A report released in Feb 2025 between Carnegie Mellon University and Microsoft Research leveraged Bloom et al’s study on the work of knowledge workers, which identifies six core components of critical thinking:

1. Gathering Knowledge - sourcing of information.

2. Comprehension - Understanding concepts, identifying themes and noticing outliers.

3. Application – Using knowledge in practical situations.

4. Analysis - Comparing and contrasting ideas.

5. Synthesis - Combining information to form new insights.

6. Evaluation - Assessing ideas based on criteria.

As we all (mostly) know, generative AI can do all these things pretty well, particularly the reasoning models.

Working cross these six components, the report studied generative AI’s impact on critical thinking among knowledge workers. Their survey of 319 professionals uncovered key findings:

• Higher confidence in AI correlates with reduced critical thinking, whereas higher self-confidence in the task at hand boosts critical thinking supported by generative AI. This suggests some knowledge workers may rely too heavily on AI without fully engaging their own minds.

• AI use shifts cognitive effort from information gathering to verification, from problem-solving to response integration, and from task execution to stewardship.

• To optimize AI’s role in knowledge work, organizations should design tools that encourage awareness, motivation, and the ability to think critically.

Trade-offs with Generative AI

Organizations implementing generative AI for productivity gains should consider any potential trade-offs. If employees, especially junior staff or new hires, lack domain expertise, they may struggle to think critically when supported with AI-generated outputs.

Some organizations promote a culture of shaping the way they approach work to drive continual innovation and industry leadership. Companies like Amazon, Microsoft, Netflix, and Google have for time encouraged a culture of promoting original thought and problem-solving, ensuring AI serves as an augmentation tool rather than a replacement for critical reasoning.

Deepseek, the AI company of recent fame based in China, and creator of the R1 reasoning model, has explicitly stated that its culture is its differentiator and moat. By prioritizing a mindset of innovation and rigorous thinking, they aim to create AI-driven solutions that maintain a competitive edge, rather than just following industry trends. This underscores how organizations that embed critical thinking into their culture are better positioned for long-term success in an AI-driven world.

Skilling beyond technology

Learning how to use the Copilots (etc) is quite clearly the first step most organisations will choose to do when they implement a generative AI solution.

What I wish to highlight though is skilling shouldn’t stop at the technology.

To assess your team’s critical thinking skills in an AI-driven environment, and what uplifts may need to occur alongside generative AI initiatives and tech training, consider the following areas:

1. Information Verification

• Do employees feel comfortable and empowered to cross-reference AI outputs with reputable sources?

• Are they skilled at assessing whether AI-generated references are legitimate?

2. Response Integration

• Do they consistently evaluate AI-generated content’s relevance to their specific needs?

• Can they refine, manipulate, and adapt AI outputs appropriately?

3. Task Stewardship

• Are they refining AI prompts to steer responses effectively?

• Do they articulate clear requirements when using AI?

4. Maintaining Foundational Skills

• Are they still practicing information gathering and problem-solving, rather than fully relying on AI?

• Do they know when they need to go beyond the information that may be presented to them, and proactively seek out other sources?

5. Awareness of Critical Thinking Opportunities

• Do they recognize when critical thinking is needed, even if AI-generated content seems polished?

6. Motivation to Think Critically

• Do they see critical thinking as essential for long-term professional growth?

• Are their targets and measures aligned appropriately to balance quality of work vs quantity of output?

7. Ability to Execute Critical Thinking

• Are they using AI features that encourage learning, such as explanations, refinement suggestions, or critiques?

• Do they know when they should be leveraging a reasoning model to support their critical thinking rather than relying on a standard LLM?

A Worked Example: Balancing AI Automation in Procurement

Consider a procurement team assessing suitable vendors. Their current process may be slow and labor-intensive, requiring employees to listen to transcripts from vendor presentations, interview key stakeholders, review proposals, and manually assess each submission before drafting responses. Generative AI presents an opportunity to streamline this process, significantly reducing the time and cost spent on vendor selection.

However, the key challenge is deciding how far automation should go. Using AI to summarize transcripts, extract key points from stakeholder interviews and follow up vendor meetings, and suggest initial decisions by creating a vendor analysis can be incredibly valuable. It allows employees to focus their efforts on more complex aspects of the process rather than spending time sifting through raw data.

On the other hand, fully automating the decision, where AI not only drafts but also decides the decision, can pose a risk. Employees may lose critical context, making it harder for them to assess the nuances of the situation or recognize deeper systemic issues in their approach to strategic vendor sourcing. Without proper oversight, this could lead to poor decision-making, a lack of accountability, and a diminished ability for employees to engage in critical thinking when making key decisions.

Striking the right balance means leveraging AI for efficiency while ensuring employees remain actively involved in evaluating responses, making judgment calls, and refining recommendations based on experience and human insight.

For every scenario, organizations should assess whether AI should act as an assistant or take over entire processes - and whether full automation risks reducing employees’ ability to think critically about the outcomes.

Recommendations for Building Generative AI Solutions

While generative AI provides substantial productivity gains, its impact on critical thinking must not be overlooked. We want to avoid scenarios where employees become over-reliant on AI tools, and experience a decline in their ability to analyze, synthesize, and evaluate information effectively. Doing so can ensure the success of these initiatives without downstream and unforeseen negative impacts.

This is particularly necessary when overlayed against your staff turnover rates, as junior employees and new team members naturally lack the domain expertise needed to assess AI-generated outputs critically until they have had more time in the job.

To ensure AI enhances - not diminishes - generative API projects, consider these steps:

• Engage domain experts to understand where AI can truly add value. Determine where generative AI is a suitable fit in these areas, and then cross reference it against what skills your people may need to develop in order to ensure the project is truly successful.

• Assess employee confidence levels. Less confident employees tend to rely on generative AI more, so focus on skill development that teaches the “why” and “how” of their roles, rather than just the “what.”

• Thoughtfully design solutions in a way that enables users to still apply their critical thinking skills. This could be done by focusing the implementation on sourcing and summarising information rather than providing recommendations. It could also include the ability for users to ask questions of the data - for example, offering a chatbot to Q&A source materials in addition to reviewing a preliminary recommendation that AI has generated..

• Be mindful of success metrics - targeting the number of issues resolved per hour in a customer service team might be very helpful, but focusing on the KPI at the expense of thoughtfully considering the needs of customers it might not be in the best interests of an organisation over time.

• And finally, don’t pause, rather identify use cases which are simple to solve, implement them, and learn from them. From there continually assess how your employees are using the tools, adjusting coaching and skilling plans as appropriate, while realising the commercial gains that generative AI brings.

Conclusion

Generative AI presents a significant opportunity for productivity gains, particularly for knowledge workers.

By fostering awareness, motivation, and the ability to critically evaluate AI outputs, businesses can ensure they harness AI effectively, without compromising essential cognitive skills that are their differentiators.