Generative AI and the Authenticity Crisis

Are We Destined To Be Prompt Engineers?

"I know your email was written with Chat GPT." Something I find myself thinking often since that fateful day November 30, 2022. Ever since Chat GPT's release, and the other generative AI tools to follow, more and more of us are using these utilities to construct our everyday correspondence and assist with basic tasks. I personally jumped in hard and fast using these tools for tasks like building starts for vague and abstract projects, especially brainstorming exercises.

Today, I find myself subscribed to GPT-4, dutifully paying my subscription each month. Most often, I use Chat GPT for sifting through work containing problems I've become blind to. In my real estate investment businesses, this could mean uncovering contradictions or bad references. In software development, I use GPT to trace hard to find bugs hiding in plain sight. Does it work? Sometimes...

I see it both ways: There are certainly times where generative tools speed things up. But there are also times when you're substituting work that should be completed by human. Correspondence being one of these instances. Two or more humans communicating among one another should not be filtered by a generative bot. Otherwise, where does this leave us as humans and as participants in production? Are we destined to live our lives plugging unrefined thoughts into chat terminals and allowing large language models (LLMs) to be our voice? Does the (very near) future consist of watching passively as two LLMs chat together? If our purpose is to serve the function of copying and pasting to and from chat terminals, that's a low value-add argument for the species.

What's the Point?

If we find ourselves observers and not primary participants, there's a fundamental issue: we're not needed. I find the idea of engineering a world of human redundancy hardly progressive. "But Chat GPT writes better and my emails seem more professional!" you counter. And maybe so. However, if a communication is being received by your peers and acquaintances, they probably have a sense of your voice. Do you expect them to believe you authored the excessively flowery and eloquent letter?

Humans have an innate ability to sense what's ingenuine. It's a threat defense. More than that, your communication demonstrates the degree of your participation. When communicating, especially through professional mediums like email, I expect much of that communication to be based around action occurring in the physical world: you would like someone to do something for you or you would like to do something for someone else (presumably for renumeration of some form). But why would I take any action if you were unable to scrape together the will to produce a complete thought in a communicable form?

Redundant, Yes. But Also Inefficient.

Emails are hard? Dear [name], I request this. Sincerely, [name]. Why is this so difficult? And it does seem increasingly difficult for many as I am receiving more and more correspondence each day without greeting or salutation. Do I need you to wish me a happy weekend? No. But at least muster the effort to identify the addressee followed by a comma and perhaps conclude your request with some conveyance of gratitude.

But this isn't a treatise on professional etiquette or a venting for lack thereof. The relief that generative AI brings us for no longer needing to gather ourselves into a state of coherence is short lived for the result is worse than original problem. Now, in your request for work product, you've given me further work by having now to sift through the unnecessary fluff of generative AI output. This is insincerity combined with inefficiency. But wasn't the point to speed things up?

The Answer: Authentic Brevity

I'll be so bold to assume I know the answer to the growing existential crises we find ourselves facing. The answer, as it happens to be, has been around for at least hundreds of years when Thomas Jefferson stated:

The most valuable of all talents is that of never using two words when one will do.

More than likely this is a misquote. Fortunately, this reframing of a Thomas Jefferson passage has likely been represented sufficiently around the internet for LLMs to have ingested and determined truth. We'll now consume this regurgitated "hallucination" from the LLMs as fact, and in their authority, believe it so. Perhaps I've digressed.

But true words nonetheless and never more applicable. Do you need to write aureate and embellished language? More than not needing to, you shouldn't. Get to the point and don't waste my time so we can get back to work. If you want to impress me with your language, use a thesaurus rather than a generative tool. If you do so then there's at least a chance of an expanded vocabulary sticking and thus the total value of human stock rising.

I did just that with aureate a few sentences back. That's the result of referencing a thesaurus in not wanted to repeat the words "eloquent" and "flowery" too often. The only person I could imagine using language like that in their natural speech would be Walter Isaacson, the Steve Jobs biographer (great books but come prepared with a pocket dictionary). But this is growth rather than inauthenticity. For every word a person speaks or writes must be written or spoken a first time. This is the distinction with generative AI: it's not your speech and it's not your writing.