Recently, I have been experimenting with various AI tools to enhance my workflow, only to reach a rather unexpected conclusion: Pen and paper remain the most effective tools for structured thinking. While AI has its merits and can undoubtedly streamline certain processes, it is far from being a universal solution. In fact, relying on AI without a proper framework for thought often results in inefficiencies and misinformation. The key takeaway from my experience is that AI should be seen as an auxiliary tool, rather than a replacement for human cognition.
The AI Tools I Tested
For this experiment, I worked with five different AI tools:
- Notta AI (for transcription)
- ChatGPT Plus (a general-purpose language model)
- Grok (X’s AI assistant)
- DeepSeek (a Chinese-developed AI model)
- Notebook LM (Google’s research-focused AI)
Each of these tools has its own strengths and weaknesses. However, what became evident is that none of them is capable of replacing the clarity of thought that comes from structured, manual reasoning.
The Limitations of AI: Hallucination and False Confidence
A major issue with AI models like ChatGPT, Grok, and DeepSeek is hallucination—the tendency to fabricate information. This is particularly problematic when dealing with qualitative analysis or abstract reasoning. When asked about the causes of a specific historical event or the factors behind a trend in financial markets, these models often generate plausible but incorrect explanations. The reason is simple: AI is not performing original reasoning; it is merely predicting statistically probable responses based on its training data.
In contrast, Notebook LM provides a much more reliable framework because it only processes information within a given dataset. This makes it particularly useful for academic research or structured Q&A scenarios where the information is clearly defined. Unlike general-purpose AI, Notebook LM does not fabricate answers—it simply acknowledges when a query falls outside its dataset.
Meanwhile, Notta AI stood out for its impressive transcription accuracy. It effectively distinguishes between different languages, making it an excellent tool for bilingual content. When I tested it on a multilingual YouTube video, it managed to separate and transcribe both English and Japanese sections with remarkable precision—something that many transcription tools struggle with.
Why Pen and Paper Still Win
Despite these AI advancements, I found myself returning to pen and paper for one fundamental reason: thinking is not just about retrieving information; it is about structuring it. AI can summarize, analyze, and even predict patterns, but it cannot replace the cognitive process of connecting ideas in a meaningful way.
When using AI, it is crucial to have a foundational understanding of the subject matter. If you lack domain expertise, you may struggle to discern whether the AI-generated response is accurate or misleading. This creates a paradox: The more knowledgeable you are, the better you can utilize AI—but if you are already well-versed in the subject, your reliance on AI naturally diminishes.
This is why I now follow a structured workflow:
- Outline ideas using pen and paper to clarify my thought process.
- Use Notta AI for speech-to-text transcription when elaborating on my notes.
- Leverage ChatGPT or Grok to refine, summarize, or supplement information.
- Validate factual information through Notebook LM, ensuring it aligns with a trusted dataset.
- Compile and finalize the output manually, maintaining intellectual rigor.
This process ensures that AI serves as an enhancement to human thought, rather than a substitute for it.
The Future of AI and Human Cognition
AI has its place in productivity, but it is not a panacea. The most effective way to use AI is not as an autonomous problem-solver, but as an assistant that automates repetitive tasks and enhances human decision-making. AI is great at processing vast amounts of data, extracting key insights, and even generating rough drafts—but it cannot replace the intuitive and creative aspects of human thought.
In an era where AI is becoming ubiquitous, critical thinking remains the most valuable skill. The ability to reason, analyze, and synthesize information is what ultimately separates human intelligence from machine-generated content. And for that, there is still no better tool than a simple pen and a blank sheet of paper.