Explainer · AI and Human Knowledge
What AI Does Not Know and Why Your Knowledge Is More Valuable Than You Think
AI has been trained on everything public, and it still has almost none of the world's real information.
AI has been trained on every single public piece of information that exists in the world, including hieroglyphs, obscure books no one remembers, and every tweet. And yet it still does not have a tiny fraction of all the information in the world. The reason: the rest lives in your brain, your customer's brain, your client's brain, your boss's brain, and your employee's brain. That unique human perspective changes every day, in every instance, in every environment, and it is exactly what both people and AI are hungry for right now.
Next step
What you will learn
- Understand the scale of AI training data and why it still leaves most knowledge uncaptured.
- Recognize that human perspective, not public data, is the largest untapped information source.
- Explain why personal and contextual knowledge changes constantly and cannot be pre-loaded into AI.
- Identify why sharing your expertise has value for both other people and AI systems.
Story sections
What AI Has Already Been Trained On
AI has been trained on every single public piece of information that exists in the world.
The starting point for understanding AI's real limits is grasping just how much it has already consumed. According to the speaker, AI has been trained on every single public piece of information that exists in the world. That list is not limited to modern content. It includes what is written in the hieroglyphs, what is in some obscure book that no one remembers, and every tweet. The speaker's summary: "It's been trained on everything."
This is not a small claim. It means that any fact, opinion, story, or idea that was ever made publicly available in written or recorded form has likely been part of the training process. Ancient inscriptions, forgotten pamphlets, viral posts, and academic papers all went in. The scale is genuinely total when it comes to the public record.
Why does this matter as a news update? Because a common assumption is that AI might be missing old information or niche topics. The speaker removes that excuse. The gap in AI knowledge is not about what was public. It is about something else entirely.
Think of the world's largest library, one that holds every book, every newspaper, every carved stone tablet, and every social media post ever made public. A single reader spent years reading all of it without stopping.
Classroom version: That reader is the AI. After finishing every shelf in that library, the reader walks outside and realizes the library held only a fraction of what people actually know, because most knowledge was never written down and put on a shelf in the first place.
Try it: List three things you know about your field or customers that you have never written down publicly. Those items are outside AI's training data by definition.
Every public fact is already inside AI. The missing information is everything that was never made public.
Why AI Still Lacks Most of the World's Information
Even after training on all public data, AI still does not have a tiny fraction of all the information in the world.
Here is the update that reframes everything: despite ingesting all public information, AI still doesn't have a fraction, a tiny fraction of all the information in the world. The speaker is precise about this. The word choice is deliberate. Not "some" information. A tiny fraction. The implication is that what AI has is almost negligible compared to the full picture.
Where is the rest? The speaker names it directly: the rest of the information in the world is in your brain, your customer's brain, your client's brain, your boss's brain, your employee's brain. This is not poetic language. It is a structural claim about where knowledge lives. The overwhelming majority of human knowledge has never been written down, published, or made accessible for training. It exists as lived experience, judgment, preference, habit, and context stored in individual people.
This has real consequences for anyone working with AI today. When AI gives an incomplete answer, it is not failing because of a bug or a missing dataset. It is failing because the information it needs exists only inside a specific person who has not yet shared it.
Imagine an expert chef who has cooked a dish ten thousand times. The published recipe is in the AI. But the chef's feel for when the pan is at exactly the right temperature, which customers hate cilantro, and what substitution works when a key ingredient runs out, none of that is in any book.
Classroom version: A sales professional's sense of which objection a particular client will raise next week, based on three years of relationship history, is the kind of knowledge the speaker is describing. It lives in that professional's brain and nowhere else.
Try it: Pick one decision you made this week based on experience or relationships rather than public information. That decision is a clear example of the knowledge the speaker is describing.
The vast majority of human knowledge lives in individual brains, not in any dataset AI can train on.
Why Human Perspective Is Unique and Constantly Changing
Each person's unique perspective shifts every day, in every instance, and in every environment.
The speaker adds a second layer to the argument. It is not just that human knowledge is missing from AI. It is that human knowledge is unique per person and changes every day for every instance in every environment. This makes it fundamentally different from static public data. A fact written in a book stays the same. A person's perspective on their market, their team, or their customer does not.
The word "instance" is important here. The speaker is not just saying perspectives change over time. They change per situation. The same person holds a different view of a problem depending on the context they are in at that moment. That level of granularity is impossible to pre-load into a model.
This is why the knowledge gap between AI and humans is not closing on its own. Even if AI ingested everything a person knew today, that snapshot would be outdated tomorrow. The moving target quality of human perspective is a feature, not a bug, and it is the core reason this kind of knowledge retains value in an AI-rich world.
A weather forecast trained on all historical weather data still cannot tell you how your neighbor, who farms the same land every year, is reading the sky this morning based on a lifetime of noticing subtle signs that never made it into a dataset.
Classroom version: A project manager's read on team morale during a specific sprint, shaped by a conversation in the hallway that morning, is a perspective that changes daily and by instance. No AI has access to it until the manager chooses to share it.
Try it: Think of a situation where your opinion on something changed in the last 30 days because of a new experience or environment. That shift is exactly the kind of live knowledge the speaker says AI cannot access.
Human perspective is not static. It shifts daily and by context, which is why it cannot be trained away.
Why Your Knowledge Is Something Both People and AI Want
People and AI both want the knowledge in human brains, because AI does not have it.
The speaker closes with a direct statement of demand: that is information that people want, and that is information AI wants, because it does not have it. This is the practical takeaway. The knowledge gap is not neutral. It creates active demand from two separate audiences at the same time.
People want it for the same reasons they have always sought out expertise: to make better decisions, solve problems faster, and understand things they cannot figure out alone. But the addition of AI as a second audience is new and significant. AI systems are actively trying to get better at tasks where they fall short, and the shortfall is exactly in the domain of personal, contextual, and experience-based knowledge that the speaker described in the previous sections.
The implication for professionals is concrete. Expertise that might have felt like background noise in a world of abundant public information now has a direct audience that is measurably hungry for it. Sharing what you know, whether with colleagues, clients, or AI tools, is no longer just helpful. It is filling a gap that cannot be filled any other way.
A local mechanic who has worked on the same model of truck for 20 years knows which part fails first in cold climates, which supplier's replacement parts last longer, and which warning sign owners always miss. That knowledge is worth more now because no manual or AI model holds it.
Classroom version: A customer success manager who knows why a specific client almost churned two years ago holds information that a new AI tool analyzing CRM data will never surface on its own. Both the client team and the AI need that context to do their jobs well.
Try it: Identify one area where your personal experience gives you a view that is not in any public document. Write two or three sentences describing that view. That is the kind of knowledge both people and AI are actively looking for.
Your knowledge is in demand precisely because AI does not have it, and that gap is not going to close on its own.
Transcript
- 0:00 When you put that together with the fact that AI has been trained on every single public piece of information that exists in the world, including what's in the hieroglyphs, what's in some obscure book that no one remembers, every tweet, everything, it's been trained on everything.
- 0:15 And it still doesn't have a fraction, a tiny fraction of all the information in the world, because the rest of the information world is in your brain, your customer's brain, your client's brain, your boss's brain, your employee's brain.
- 0:26 It's there, their unique perspective, and that unique perspective changes every day for every instance in every environment.
- 0:33 And that's information that people want, and that's information AI wants, because it doesn't have it.
Questions
If AI has read everything public, why does it still get facts wrong?
The speaker's point is about the scope of what AI has trained on, not about accuracy. Training on public data does not guarantee correct recall or reasoning. The more relevant issue raised is that AI lacks the private, contextual, and experience-based knowledge that was never made public, which leads to gaps rather than errors in a narrow factual sense.
Does this mean AI will eventually learn from people's private knowledge too?
The speaker describes the current state: AI does not have this information now. The reason it lacks it is structural. Human perspective changes every day in every instance and every environment, so even if AI gathered more private knowledge, the moving-target nature of human perspective means the gap would persist. The point is that human knowledge has ongoing value, not a fixed endpoint.
Who specifically is affected by this knowledge gap?
The speaker names several groups directly: you, your customers, your clients, your boss, and your employees. Anyone whose expertise, relationships, and lived experience inform decisions that AI tools are now being asked to support is sitting on information that both people and AI are actively looking for.
What should I actually do with this information?
The speaker's framing points toward actively sharing expertise rather than assuming AI or published resources cover it. Whether that means documenting institutional knowledge, contributing context to AI tools you use, or making tacit expertise more visible to colleagues, the gap is only filled when the people who hold the knowledge choose to share it.
Glossary
- Training data
- The information an AI model learns from before it is deployed. The speaker states this includes every single public piece of information in the world, from hieroglyphs to tweets.
- Tacit knowledge
- Knowledge that lives in a person's brain based on experience, judgment, and context, and that has never been written down or made public. This is the category of knowledge the speaker says AI still does not have.
- Unique perspective
- The speaker's term for the individual viewpoint each person holds, which changes every day for every instance in every environment and cannot be pre-loaded into any AI model.
- Knowledge gap
- The difference between what AI has trained on (all public information) and what it still lacks (the vast majority of human knowledge stored in individual brains).
Resources
- Micro-Learn Library Explore more short lessons on working with AI, building on the concepts introduced here.
- How to Work With AI Effectively A practical next step for anyone who wants to close the knowledge gap the speaker describes by contributing what they know to AI-assisted workflows.