Our AI invented the numbers. We caught it.
The hosts discuss the importance of double-checking AI outputs and validating information, emphasizing that this is not a new problem but rather a longstanding practice required with any information source. They explore how AI tools can be customized to individual workflows and contexts, arguing that understanding foundational methodology matters more than specific tools.
Summary
The conversation begins with the hosts acknowledging that AI can produce false information, but frame this as an existing challenge rather than a unique AI problem. They stress the importance of verification practices that have always been necessary in academic research, journalism, and professional work. The discussion then pivots to the practical application of AI in productivity workflows, with both hosts sharing examples of how they've integrated AI into their content creation and article writing processes. Rather than using AI as a push-button solution, they emphasize the importance of iteration, customization, and building systems that align with individual expertise and cognitive styles. A significant portion of the conversation focuses on the concept of context and how AI systems become powerful when they have access to complete, well-structured personal context and information. The hosts argue against tool dependency, advocating instead for local control of data while leveraging powerful external services. They introduce the ICO methodology as a foundational, tool-agnostic framework that remains constant even as specific technologies change. The conversation emphasizes that successful AI implementation requires understanding core principles rather than chasing trending tools, and that building effective systems requires sustained focus on specific use cases rather than broad experimentation. Finally, they discuss momentum and compounding effects, noting that their membership growth from 1,000 to 7,000 people has reinforced their methodology while remaining fundamentally unchanged over five years.
Key Insights
- The problem of AI producing false information is not unique to AI—it's a longstanding issue that existed with Wikipedia, word-of-mouth information, and any unverified source, making verification a fundamental practice across domains rather than an AI-specific concern.
- The real power of AI systems comes from having complete, well-structured personal context available—something that has never been possible before in computing history, enabling tools to naturally understand and respond to user needs without extensive explicit instructions.
- Success with AI requires understanding foundational, tool-agnostic principles and methodology rather than becoming dependent on specific tools or following trends, as these core principles remain constant even when underlying technology changes.
- Building effective AI systems requires sustained focus on one specific high-value use case for extended periods (weeks to months) rather than broad experimentation, with parallel testing against existing systems to avoid abandoning proven workflows prematurely.
- The distinction between privacy and dependency is important—one can safely use cloud services without being dependent on them if data can be fully extracted and migrated, maintaining autonomy even while trusting external services.
Topics
Transcript
[0:00] All right, everyone. Welcome back to another session this week with my dear friend Paco. Welcome, Paco. Today we talk about >> well, who would have guessed AI and in the realm of productivity. The thing is I just published a video on YouTube on the main channel um talking about doing research with AI and I think that's something we should talk today about. [0:30] >> Um the video is all about double-checking what you actually get back from AI and this video it was funny enough uh the research for the video itself produced fake numbers for this video. So it is included in the video itself. >> But this is what I did from the get-…
Full transcript available for MurmurCast members
Sign Up to AccessMore from ICOR with Tom | AI Productivity
An AI Just Handed Me a Fake $67B Statistic
The speaker shares how an AI confidently generated a completely fabricated $67.4 billion statistic while researching AI hallucinations, then explains his systematic approach to combating AI misinformation: using dual independent search engines to cross-check answers and identify where hallucinations occur.
AI Is Quietly Taking Your Memory
AI companies are racing to own users' context and memory through integrated tools like Claude in Slack and ChatGPT's auto-memory features, creating a lock-in trap. The speaker argues users should own their context (expensive, irreplaceable) in local plain text files while renting the AI model (cheap, replaceable), and demonstrates his personal knowledge assistant system built on markdown files that work with any AI model.
My AI Team Now Has an Interface. All 12 Agents. Free.
Tom, creator of the iCola methodology, releases My PKA version 3 with a free interface (My PKA Cockpit) that provides 12 pre-built AI agents for productivity, note-taking, health tracking, document management, and task planning—all running locally in a folder structure accessible via Claude and a web interface.
They Restricted Fable. I Switched Models in One Folder.
The creator explains why the removal of Claude Fable didn't disrupt his workflow, arguing that model-agnostic local folder setups with structured instructions make users resilient to AI changes. He uses an analogy of a dark room with a chair to explain how context and instructions guide AI behavior. The core message is that building robust local documentation and agent orchestration systems matters more than chasing the latest AI model.
Claude just killed ALL Note-Taking, Planner, and Health Apps. Here is proof.
The creator demonstrates a custom-built, locally-run personal knowledge management interface powered by Claude AI that replaces multiple productivity apps including note-taking tools, planners, and health trackers. The entire system lives in a single local markdown folder, requiring no subscriptions or cloud dependencies beyond a Claude plan. The interface will be made available for free to members of their myICOR community.