A few days ago, I read a newsletter by Mr. Thai, aka 'The Hacker Engineer,' titled "Five Levels of Ignorance in the AI Era." I'll quote a passage and then share a story below.
[...] the "five levels of ignorance" by Ngo Quang Hung, citing Phillip G. Armour (OI stands for Order of Ignorance):
4OI - Meta-Ignorance: I loosely translate this from the term 'meta-ignorance.' At this level, you don't know anything about the five levels of ignorance.
3OI - Lack of Process: At this level, you lack a specific process to discover that you don't know that you don't know about X. In other words, at this level, you don't know how to find the things you don't know that you don't know 🙂.
2OI - Lack of Awareness: At this level, you don't know what you don't know about X. Obviously, we can't give an example of 2OI! However, sometimes reading books, newspapers, or computer science blogs (!), I can find many things I never knew I didn't know, and thus I move those things up to 1OI. While I am generally at 1OI with quantum mechanics, there are certainly specific objects within quantum mechanics where I am at the 2OI level.
1OI - Lack of Knowledge: To... achieve this level of ignorance, you must know that you lack knowledge about X. Example: I know that I don't know anything about quantum mechanics. Reaching this level is already good, because if needed, I can seek out books and resources on quantum mechanics to learn more.
0OI - No Ignorance: To achieve this level, you must know X and be able to prove that you know X. Example: I know how to write a blog!
ChatGPT makes everyone think they are at the 0OI level: everything has a solution, and it reads very smoothly. Everyone thinks they've understood, but in fact, they've only understood the reading. [...].
Let me tell you about the assignments in my Strategic & IMC Planning class, which are quite demanding (in a positive way). Students need to complete a part of the homework every 3 days for me to review. For example, in the target audience analysis section, students need to do consumer segmentation, and I require them to interview people to complete the assignment and derive the main insight.
During the review session, I immediately caught the students who hadn't conducted interviews and had used AI, because the result read very "smoothly": the personas were named beautifully, every aspect of a persona seemed logically sound, and the wording was too fancy. However, after reading it, I didn't feel like this was... a Vietnamese person (or if it was, perhaps a global citizen), and the main insight derived was too generic and non-actionable (not useful).
I don't know about other industries, but "reads smoothly" is the sweet trap that makes planners—or clients/agencies using AI to replace planners—think they have grasped and found a cool answer to a brief (which might also have been written by AI).
I figured this out because I've personally tried to replace myself with AI dozens of times and failed. I'm also a co-founder of an AI startup, and most importantly, I've spent 12 years achieving the "0OI - No Ignorance" level in planning.
There is no shortcut to climb from 4OI to 0OI. If there is, it's AI fooling you.
So, Planners, what level of ignorance do you think you've reached in the age of AI?
P.s: I'm not anti-AI. When teaching, I guide students on which parts AI can SUPPORT them effectively, but it cannot do your thinking for you.
