Predicting what's inside

[fluffy]

I've been looking into research on mbti and data collection. So far a study I read states 20% accuracy on dichotomy (I/E)(N/S)(T/F)(J/P)

The study was only a small one with no long term research done on functions.

This made me think.

Text analysis has existed a long time.
With LLM this has become even more refined.

IBM watson in 2011 was 100 trillion calculations a second, today a laptop can do this many and has way better algorithms.

I believe that as work has progressed we could find a greater accuracy on mbti with enough data on a person. Tho this would need to be input into the machine from different sources. The result would better account for ones type along with other characteristics. We may be able to duplicate a person's personality very accurately as well.

Are other people interested in this research?

What could we know about people from this?

My guess is that the inside would be different from the outside is some instances. Like how we represent ourselves with different avatars.

 

[dr froyd]

i put your post into an LLM, it said youre intp

Based on the writing style and content, I would guess that the person who wrote this is an INTP (Introverted, Intuitive, Thinking, Perceiving) type. Here's why:

• The writer is interested in researching and analyzing data, which is a common trait among INTPs who enjoy exploring abstract ideas and concepts.
• The text is written in a logical and analytical tone, with the writer presenting their thoughts and ideas in a step-by-step manner, which is consistent with the Thinking (T) aspect of the INTP type.
• The writer is also open to exploring new ideas and possibilities, such as the potential for text analysis to improve MBTI accuracy, which is a characteristic of Intuitive (N) types who enjoy exploring possibilities and imagining alternative scenarios.
• The writer's use of phrases like "This made me think" and "My guess is" suggests a somewhat casual and speculative approach to ideas, which is consistent with the Perceiving (P) aspect of the INTP type.
• The writer's interest in understanding people's personalities and characteristics, as well as their potential to present themselves differently in different contexts, suggests a curiosity about human behavior and a desire to understand underlying patterns and motivations, which is also consistent with the INTP type

 

[fluffy]

That was kinda the point of how a.i. doesn't understand the difference between functions and acronym dichotomies.

What makes what I said Ti rather than Te

Why are possibilities Ne and not Ni

Chatgpt I have used in the past but it's not the best at reasoning it's way to a conclusion. It often tells you what you want to hear.

I once asked it to calculate how many miles the earth rotates in 5 milliseconds and I had to redo everything for it. It's not exactly rational.

It could be that IBM system would do better but I don't have the money. I have gigabytes of data but it can take time to understand what these systems are actually doing. The staff at IBM understand the bias that counts as they actually trying to make money for customers where ambiguous marketing doesn't get results desired. The short piece of text is not going to tell you much about individuals where usually you need hundreds of data points connected in a network like at Facebook or Googles knowledge graph.

Still it's not exactly the fault of any particular a.i. to get things wrong because context windows are short. The more data you have and how it's structured will increase the accuracy of predicting someone's psychology if the analysis tools are contructed in the right way.