For a while, I have been thinking about how we know and learn things, especially in today’s world where we are constantly bombarded by information in the attention economy, told that search and AI will make us dumber, and the “algorithm” encourages us to participate and debate (put politely) on varying topics with people across the world. In thinking about how we know and learn things, I find that we (or at least, I) go through these phases:

• Unknown unknown: this is an area where we don’t know that this knowledge or concept exists – most of the world’s knowledge exists here for any given individual.
• Known unknown: this is when we know that the area exists but we don’t know anything or much about it.
• Awareness: this is when we know about the area, and have shallow knowledge about it. We may have read a bit about it but we should resist the urge to comment strongly in this area without doing further research.
• Knowledgeable: finally, we know enough to have an opinion formed out of experience or research. We should be able to participate in conversations and debates.
• Deeply Knowledgeable: in my mind, this comes from being aware of the nuances and context of the knowledge we possess, and are able to “see the other person’s view” when in a debate or discussion about it.

I prefer not to use words like Expert or Specialist since people have meanings attached to those words. As we think through this a bit more, I would say that there are three sub-levels that kick in, especially with skills:

• Knowing how: these are gains in competency as we move from “Awareness” → “Knowledgeable” → “Deeply Knowledgeable”.
• Knowing when: this probably exists in the “Knowledgeable” → “Deeply Knowledgeable” phases and allows us to decide when it’s a good idea to use what we know.
• Knowing when not: once we get really good with something, we start to think that it is the solution to all problems. The next level of awareness is to be able to decide when this is not the real or best solution and we should use something else. Funnily enough, when I wrote this initially, I was thinking about programming and such, but I think it also applies to things like data – as you get better, you (should) know when your data actually supports your argument and (almost more importantly) when it doesn’t.

In today’s world, there is quite a debate about using AI and if it means the end of search. So, how do we acquire knowledge and bring things from outside of our knowledge universe into a place where we are deeply knowledgeable about it? I think different approaches work better at different points:

• Meeting and talking to people or using AI is a great way to discover an unknown unknown, and to bring it into your knowledge universe. You could use AI (or specifically a large language model) to question (like you would a human): What is an approach that an expert in this area might take? What are different things that would help me have a wider understanding of this area? and you’re likely to come across new ideas, frameworks, and concepts that you should know about.
• Traditional web search (for me) works better when something is in the Known Unknown or Awareness areas. Search up things, find differing viewpoints, reason, and try to form an opinion and holistic understanding. For now, I find that LLMs are too prescriptive and final in their summarised opinion. This may be great to have a talking point or just enough to say “I don’t know enough” but I feel that exposure to multiple sources of information is more useful on your path to become knowledgeable. This is why hands-on experience (or personally done primary research) will also add significant value.
• When you’re knowledgeable or deeply knowledgeable, I think all forms of exposure have value. When you’re knowledgeable, you can also use LLMs to probe deeper into the corners to see if there are things hiding there. If you know enough to form hypotheses, you might also be able to ask the non-obvious questions: According to experts, what is the role of XYZ in this area? or When is it recommended to use which method? and so on. You might get this knowledge also from books, blog posts, and conversations, but it is likely that an LLM might be able to help with summarising multiple sources that you did not even know existed.

That’s my brain dump on this! Thank you for reading this far. If you have some comments, I’d love to hear from you. Feel free to connect or share the post (you can tag me as @onghu on X or on Mastodon as @onghu@ruby.social or @onghu.com on Bluesky to discuss more).