You're Applying to a Job Right Now and You Didn't Even Know It
Try this 2 minute exercise on GPT - this could be the future of job allocation
There are two prompts I’d like you to try that will tell you a lot about yourself and how you stack in comparison to others.
Prompt 1 (How do you talk to me vs other people):
I know that ChatGPT learns the user’s behaviors from many conversations and adjusts accordingly. What have you learned from my behavior and how have you adjusted your responses in the way you respond to me, compared to your baseline and/or how you respond to other people, on average and in general?
Watch what GPT tells you about the style of communication it uses in order to tailor to your thinking and writing style. Here’s mine:
Ok, next. Try this simple prompt.
Prompt 2 (Percentile Comparison)
Compare me to other users in terms of percentile for depth, breadth, etc….
Here’s what I got:
Ok, cool. Why am I sharing this with you?
Do you see what’s happening here?
AI can—in a manner more intimate and unbiased than anyone who has ever dealt with you—map and understand the quality and patterns of your thinking, profiling your strengths and weaknesses in a way that no existing standardized test could ever come close.
Imagine what this could entail as far as job hiring and allocation go (assuming jobs exist as a normative concept 10 years from now).
A company collaborates with OpenAI and asks for your permission to let ChatGPT share with them a profile of who you are and sends them something along the lines of what I pasted above.
On the one hand, that’s kind of cool… a maximally-efficient mechanism that aims towards a perfect meritocracy?
On the other hand, your private conversations… the mirrors of your soul, as it were, were exploited and used to serve someone else’s instrumentality.
Now, imagine that we have a centralized or totalitarian government whose leadership decides to allocate every individual to some job that the AI says he or she is best equipped for. The AI has modeled your behavior to an unparalleled level of precision and will ultimately decide whether if your best contribution to society is through being scientist or a human battery.
Of course, if either of these situations (both of which entail OpenAI selling models of your behavior to other institutions) occur, their biggest impact might be only in the first generation, where a person’s private interactions with an AI are authentic, since the later generations will begin to game the system by asking contrived questions that they hope give a certain impression.
i.e. The future student can plan: *Let me think about the kind of question someone with the capacity of becoming a doctor would ask the AI, so that the AI can infer I’m the kind of person who could become a doctor*
I mean, heck, the student could ask some other localized AI to generate the type of question that would signal the kind of traits they hope to be known for.
Though even there, you would think that the AI could detect disingenuous attempts to game it.
At the same time… if someone were smart enough to fool AI, then maybe they would deserve to be considered for any kind of future job track.
Who knows…
But it is interesting. And the market for this kind of exploitation would be MASSIVE.
Let’s see what happens.
In the meantime, why don’t I create a job posting to hire someone who can talk to me in the way that I apparently like the most. 😄
Looking for a teammate who communicates in an analytical yet collaborative tone and provides a multi-layered analysis with philosophical and technical precision through multi-turn iteration. . .
—Drago