EMMW

Beware: Tunnel Ahead

25 April 2026·
words
·5 min read
Energy
Alignment
Operations
Wonder

Summary

How can we continue to expand our thinking and understanding when AI stewards are guiding us all down the same predictable path?

Imagine a library ... what do you see? Walls of elegant shelves reaching to the ceiling, open accessible, quiet? A space without expectation with a friendly assistant to guide your exploration. A library is the definition of expansion.

The internet was built to be an endless library. Search was the first iteration of the helpful assistant, now we have AI. But rather than expanding - all I feel is contraction.

AI is a predictability engine providing the most probable outcome. It struggles to provide the unexpected or the challenging because that is not what it was designed to do! LLMs are rewarded for giving you the right answer, where 'right' was defined by one specific culture and viewpoint.

The trend I'm seeing is that the search engines (our traditional portal to the internet) is being replaced by AI to act as stewards of the information. If we are not careful we will find all of our thinking guided - no directed - down a series common of pathways. Accidentally (or intentionally) blocking out the new, the unexpected, the 'unconventional'. (Yes I know search is also biased but at least it was all there if you looked for it).

How can we continue to expand when we are being ushered down the predictable path?

Maybe a more powerful (and personal) question is how do you want to interact with information, technology, AI, and the internet? There is no right answer but you need to have the awareness to make a choice.

--

Examples of Tunnel Vision

AI Meeting Summarization

Rather than writing our own minutes (or summaries) of meetings we rely on AI to 'take notes' for us. Is anyone else finding it weird there are more AI note-takers in the meetings than people anymore? AI can transcribe the words, but not the emotion and the energy of the meeting. Since it is a probability machine - it will summarize the bulk of the conversation - but will it capture & highlight the ah-ha moment - the breakthrough that happened in the final moments of the conversation that will act as the new baseline?

Podcast Snippets

I was exploring various tools to capture a snippet of a podcast that triggered something for me. I simply wanted to capture the words at that moment - to come back and riff on later when I had a moment to sit and reflect. Sometimes I would even leave myself a voice note about it ... a simple ask no?

Well much to my surprise I found that all the podcast capture tools were already embedded with AI. So rather than just giving me the raw words from my snippet. It provided me a 'helpful' nterpretation of what was said in this section - it told me what to think about this section rather than allowing me to think and share my own insights. Not cool!

Book Reports

In high school (late 90s in the USA) I remember writing a book report for my AP English class. Over the course of the year I had to pick one author and read 10+ of their works and then write a report exploring a theme. I chose Daniel Quinn and his opinion about religion ... I'm sure I could go out to an AI agent today, pick 10 books by Daniel Quinn and tell them to give me an overview of his take on religion. I could steer the perspective and the voice but I would get a summary of what everyone else saw in the patterns and have published online. I would get an integration of how we study and explore religious texts - but I would lose the opportunity for my voice to expand, to explore new perspective and to share my unique perspective in the world.

Health

If I ask 'head cold remedies, holistic' or better yet 'signs of burnout' - I will get the same old tropes even if I ask for alternative approaches. In the burnout land, you will get the same list of symptoms and approaches that are generic and bland - separated from your experience and actual needs. The AI might 'speak' to you like your unique situation matters and is considered but mostly you will get the same general solution.

Construction or Baking, etc

If you ask AI for help on how to wire a light switch - you will likely get a generic American solution rather than something that is specific for your region and infrastructure. And if you follow blindly assuming that the system has your best interest in mind (haha) - you might end up in danger.

Honestly the last one is a guess but it would not suprise me - and you get the point. Maybe it is baking or pottery, or car engines - what are we missing out on by assuming it we are getting the 'complete' picture.


In this era it will be more important than ever to question the source of your information and learn how to challenge and verify the results ... but how can you do that when you no longer have access to the library?

Now is the perfect moment to define how we want to work with technology/AI & how we want to be supported by it before we forget there is a world outside the tunnel we are 'asked' to stay within.

Join the Conversation

AI Support

ContentNone
Support
  • SEO/AEO

History

Created 25 April 2026Evelyn Pacitti

raw thoughts on AI tunnel vision

Updated 25 April 2026Claude · Opus 4.6

SEO/AEO generated