As tech professionals, we’re constantly navigating a landscape shaped by rapid advancements. Generative AI, tools like ChatGPT, DALL-E, and Copilot, have quickly become indispensable in our workflows—boosting productivity, simplifying repetitive tasks, and enabling faster problem-solving. But with convenience comes an important question: how does generative AI impact our critical thinking?

A recent study by Lee et al. (2025) provides a fascinating glimpse into this intersection. The study surveyed 319 knowledge workers across industries to understand how generative AI affects their critical thinking. The findings? Generative AI doesn’t just make work easier; it also shifts the cognitive landscape in profound ways. Let’s dive deeper.

What is Critical Thinking, and Why Does It Matter?

Critical thinking is more than just “being logical”—it’s about analyzing, synthesizing, and evaluating information to make informed decisions. Lee et al. used Bloom’s taxonomy to frame critical thinking, breaking it down into six key activities: remembering facts, understanding and organizing ideas, solving problems, analyzing issues, creating new ideas, and evaluating quality.

Below is a visual representation of Bloom’s Taxonomy and how each cognitive skill maps to generative AI workflows, such as debugging systems, designing solutions, or reviewing outputs:

In the world of tech, critical thinking manifests in debugging complex systems, designing scalable architectures, or evaluating AI-generated outputs. It’s not just a skill; it’s a mindset that ensures we don’t just rely on tools but use them effectively to elevate our work.

In the world of tech, critical thinking manifests in debugging complex systems, designing scalable architectures, or evaluating AI-generated outputs. It’s not just a skill; it’s a mindset that ensures we don’t just rely on tools but use them effectively to elevate our work.

The Double-Edged Sword of Generative AI

Generative AI offers significant benefits, but it also introduces new challenges that require thoughtful navigation.

On the positive side, AI tools have brought about dramatic efficiency gains. They automate repetitive and mundane tasks, freeing professionals to focus on higher-impact activities. For instance, GitHub Copilot can generate boilerplate code within seconds, allowing developers to dedicate more time to designing and refining software architectures. Similarly, tools like ChatGPT can draft initial versions of documents, enabling writers and engineers to begin with a strong foundation.

Moreover, generative AI serves as a powerful scaffolding tool for complex workflows. By organizing and presenting information in a coherent manner, these tools simplify processes that might otherwise require hours of manual effort. For example, a developer tasked with writing user documentation can use AI to generate a structured outline, saving valuable time and energy. The ability to instantly retrieve and organize data enables a level of productivity that was previously unattainable.

However, the convenience of AI comes with risks. Over-reliance on generative AI can erode critical thinking skills. When professionals defer too much to AI outputs, they risk losing the depth of understanding required to challenge or improve upon those outputs. For example, using an AI debugger without investigating the underlying problem might resolve an issue in the short term but leaves knowledge gaps that could prove costly later.

Another challenge lies in the shift from problem-solving to verification. AI tools excel at generating answers, but it’s up to humans to ensure those answers are accurate and contextually appropriate. This verification process often requires a level of expertise and diligence that users may not always apply. In high-stakes scenarios, such as crafting legal documents or analyzing financial data, unchecked AI outputs can lead to significant errors.

Finally, the study highlighted the phenomenon of over-trust in AI. Professionals with high confidence in AI capabilities tend to engage in less critical evaluation of its outputs. While this trust can streamline workflows, it also opens the door to mistakes when AI-generated solutions are incomplete or inaccurate. Striking a balance between trust and scrutiny is essential.

How Generative AI Changes the Thinking Game

The study revealed several profound shifts in how tech professionals approach tasks in the age of generative AI. These shifts redefine our roles and responsibilities in workflows.

First, there is a noticeable shift from gathering information to verifying it. Previously, much of the effort in knowledge work revolved around collecting and synthesizing data from various sources. With AI tools like ChatGPT, the heavy lifting of data retrieval is largely automated. However, this convenience places the onus on professionals to validate the information provided. Whether it’s checking the accuracy of a generated report or ensuring compliance with industry standards, verification has become a critical component of the workflow.

Below is a visual comparison of cognitive effort across tasks before and after the integration of generative AI, highlighting how these shifts manifest in practice:

Second, problem-solving has evolved into response integration. Generative AI often provides ready-made solutions, but these solutions are rarely plug-and-play. For example, an AI-generated SQL query might need to be adapted to fit the specific schema of a database. This process of tailoring and integrating AI outputs into broader systems requires a nuanced understanding of both the task and the context.

Finally, the role of professionals has shifted from task execution to task stewardship. Rather than performing tasks directly, many professionals now guide AI tools to achieve desired outcomes. This involves refining prompts, steering outputs toward specific goals, and ensuring that the results align with quality standards. While this shift enables greater efficiency, it also requires a heightened sense of accountability, as the ultimate responsibility for the work still rests with the human user.

Second, problem-solving has evolved into response integration. Generative AI often provides ready-made solutions, but these solutions are rarely plug-and-play. For example, an AI-generated SQL query might need to be adapted to fit the specific schema of a database. This process of tailoring and integrating AI outputs into broader systems requires a nuanced understanding of both the task and the context.

Finally, the role of professionals has shifted from task execution to task stewardship. Rather than performing tasks directly, many professionals now guide AI tools to achieve desired outcomes. This involves refining prompts, steering outputs toward specific goals, and ensuring that the results align with quality standards. While this shift enables greater efficiency, it also requires a heightened sense of accountability, as the ultimate responsibility for the work still rests with the human user.

Practical Tips for Critical Thinking in an AI-Driven World

To thrive in this evolving landscape, tech professionals must adapt their approach to critical thinking. Here are some strategies:

Cross-verification of AI outputs is essential. Don’t assume that AI-generated content is accurate or complete. Whether it’s a piece of code, a design mockup, or a report, take the time to verify its validity against trusted sources or your own expertise. This step ensures that the outputs meet both the technical and contextual requirements of the task.

It’s also important to use AI as a partner rather than a replacement. AI can assist in brainstorming ideas or drafting initial frameworks, but the human touch is crucial for refining and enhancing the results. By actively engaging with AI outputs, you maintain control over the creative and analytical aspects of your work.

Additionally, cultivate curiosity and skepticism. Challenge the assumptions behind AI-generated outputs and explore alternative solutions. This habit not only sharpens your analytical skills but also ensures that you remain actively involved in the problem-solving process.

Lastly, establish clear quality standards. Define the criteria by which you evaluate AI-generated work, such as clarity, efficiency, and scalability. Having a structured framework for evaluation helps maintain consistency and ensures that the final product meets professional standards.

The Path Forward: Designing AI for Critical Thinking

The responsibility for fostering critical thinking doesn’t rest solely with users. Developers of AI tools must prioritize features that encourage reflective and deliberate use. For instance, incorporating source verification prompts or providing alternative suggestions can nudge users toward deeper engagement with the outputs.

Training programs focused on AI literacy can also play a crucial role. Workshops that teach professionals how to effectively interact with AI tools can bridge the gap between convenience and critical thinking. These programs can help users develop the skills needed to evaluate and refine AI-generated outputs effectively.

Finally, embedding feedback mechanisms within AI tools can enhance their utility. Features that explain the reasoning behind AI outputs or highlight potential areas for improvement empower users to make more informed decisions. By fostering a collaborative relationship between humans and AI, we can ensure that these tools augment rather than undermine critical thinking.

Conclusion: Striking the Balance

Generative AI is here to stay, reshaping how tech professionals work and think. While it unlocks incredible efficiency and scalability, it also challenges us to stay critical, curious, and engaged. By embracing AI thoughtfully, we can elevate both our workflows and our skills—ensuring that the tools we build don’t build complacency in us.

Citation: Lee, H.-P., Sarkar, A., Tankelevitch, L., et al. (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. CHI Conference on Human Factors in Computing Systems (CHI ’25). https://doi.org/10.1145/3706598.3713778.

Disclaimer: This article was created with the assistance of AI tools to enhance clarity and streamline content creation. All final edits and perspectives are original and my own.