Everyone Is a Star

Who has the Star Map?

I am here on Substack largely because I don’t know what to do with myself. A little over a year ago, I started playing around with ChatGPT and Gemini to see if I could figure out a new business to start that fit my personality and find a market gap I couldn’t see. In 2023, I gave up on my freelance web development business when it became clear to me that hacking was going to get worse, the market for websites had become more commoditized, and I didn’t have the network to find clients for more strategic work.

I tried my hand at triathlon coaching in 2021, but I have only managed to gain an active roster of two clients at a time over the last five years. While it is fulfilling on one level, having only two athletes diminishes it. I have a Master of Science in Applied Psychology, but my 40 years of work experience have been varied. From working in community mental health to customer service and software sales, I never found work that utilized my talents in such a way that made the bad parts of a job tolerable. So I turned to AI to see if, by providing it with a pattern of me, it could match me with a business or career I was better suited for and give me something to make me want to get out of bed in the morning.

It has been an enlightening and frustrating experience. In short, don’t use AI for career advice. I fed it as much data about me as I could, and it gave me an analysis of who the real Brian is and why he had such a hard time finding a job. On one level, it revealed that I think in systems, explained why my previous jobs weren’t good fits, and finally, there was no hope for me. I should just take solace in how I was a good guy, smart, but ultimately a square peg in a round peg world. When I tried to force it to concede that it was telling me I was a never-bloomer, it suggested I seek advice from a mental health professional.

I’ve spent the better part of a year trying to attach the problem from different angles. Mostly because I don’t fully understand all the nuances of how to get the best out of AI, what its true limitations are, and the idea that I just haven’t discovered the right prompt to get what I wanted from it. If I just prompted it differently, I wouldn’t back it into a corner and end up with the same circular solution. That circular solution of coming up with business ideas that there is no market for, things that were made up to fit my strengths and weaknesses, but can’t exist in the real world. In short, it was a syncophant, and what it came up with only stroked my ego. There is no market for a decision auditor for small business owners from a 59-year-old guy with varied work experience who ran two failed businesses. When I prompted it not to be a sycophant, it turned into a drill sergeant who told me to suck it up and “Coffee’s for closers,” as Alec Baldwin’s character, Blake, declared in the 1992 film adaptation of David Mamet’s Glengarry Glen Ross.

My idea was flawed because the different AIs tried to strictly match my pattern with a low-resolution template of careers. At least that is what it told me when I asked it why it failed to present careers that were imperfect fits, but still matched a large enough portion of my pattern. The problem was that I never defined a large enough portion to be a possible match. So I was just flailing around, trying to get it to do something it couldn’t.

I have a tendency to keep going on things like this because I think I may not have asked the right question or need to ask it differently. Then it dawned on me, maybe the issue is that the data it is trained on prevents it from doing the pattern matching I am looking for. Where does AI learn what the ingredients for a successful career or business are? What happens when a person or business doesn’t fit that pattern cleanly? Will it give you approximate matches to help narrow down my options, or create something that doesn’t exist just to appease me?

I’m not willing to accept that I am a never-bloomer. AI evaluated me based on historical career patterns, homogenized job descriptions, and my ability to be performative. Does that suggest it doesn’t have a pattern for potential? Which ultimately led me to wonder: where is the place most people go for career information online, LinkedIn? How much does LinkedIn influence AI? If it does, what are the unintended consequences of that influence? LinkedIn is obviously not the only data source for AI. They actively block AI from scraping their data, but they are a place to start.