Laura Bungarz

I've used AI to aid my writing process. I wrote my entire philosophical framework with AI as a thinking partner.

After probably 75 000 messages over two years or so, there are no more em dashes or phrases using the word “gently” in the output. ChatGPT now closely mimics my voice and how I talk. It turns out, you can make the em dashes go away without repeatedly prompting for it. Context and history play an important role in how AI responds to you.

Why did I use AI to write the framework? Because I wanted an a-moral, non-emotional, reasonably neutral framework that explained cause and effect independently of human story and perception. I used AI to help examine situations logically and question the assumptions behind them.

A simple example was a tree falling on my house. How does human perception work in that situation? What is actually true? What is potentially true? What is mostly human perception? What are the possible outcomes? What is the person reacting to? Perception or reality? I also used historical references. Given what we know about what happened, was the outcome logical within the context of the framework? Plug the World Wars into it. Does it make sense? I did this with ChatGPT and every other model I had access to. I'd feed their outputs back into ChatGPT, too. I wanted the AIs to break each other's logic. Agreement between models wasn’t enough. I wanted to understand whether the explanation held up and where it fell apart.

Across thousands and thousands of messages, with a lot of iteration, discussion, questioning and back and forth, I ended up with the framework I wanted.

Through that process, I learned a lot about prompting and getting AI to produce what I needed. I also began to understand the limitations of different models and which were more suited to working within a morally neutral framework. Some repeatedly brought moral framing back into hypothetical and historical discussions. Others produced output that was more machine readable than human readable, making them difficult to work with.

Now, ChatGPT is Grammarly with a personality. It's really good at cleaning up my writing after I write something. But I didn't get that from a good prompt. I got that from extensive history and context.

AI in writing is frustrating on a cold AI. You get the standard output format, complete with em dashes and “gentle” phrasing. That was true for me in the beginning, too. It took a long time to get ChatGPT out of therapy mode and get it to stop dragging morality back into things. But with enough effort, you can get customized output that closely reflects how you talk and follows how you think.

The average user isn't going to work that hard. I get that. I also believe that over time through training, the standard AI output will improve. For now though, persistence and time are the fixes that make AI output more helpful than not.

That’s not to say that AI is infallible. It does make mistakes. It is not perfect. You have to be willing to question its reasoning, point out the mistakes, and do the work.

From my perspective ChatGPT and any AI, can help to carry some of the mental load of organizing thoughts and generally structuring your writing. It can also help you find holes in your own thinking, show you where there might be a gap in your writing, or provide additional prompts that would add value to your work.

I understand that there are people out there that will simply prompt the AI, copy the output, and paste it as though they wrote it. When photocopiers were first available publicly, I’m sure somebody spent time photocopying entire books. Frankly, you can still check out a book from the public library or even buy it at the store, take it home and scan every page if you really want, and then bring it back. AI doesn’t change that, it just exposes that the possibility to make an electronic copy of something and repurpose, reorganize, or repackage it has been around for a long time.

Technology in general has changed how we function as a society. Music artists are having to figure out streaming. How do you make money in a world where everything is downloaded and records or CD’s are not the primary medium for sales? Artists were always at risk for theft. How many of us spent time listening to the radio with a blank cassette tape in the boom box ready to record a song when it came on?

Theft got easier, but we never actually resolved the reason for the underlying theft. Why did we record songs off the radio? Why do people try to find music for free? Why would somebody photocopy an entire book or take a book home, scan it, and return it? What is the underlying reason for that? Solve the original problem instead of just trying to put enough obstacles in the way to prevent it.

Why do people try to find things for free? What parts of society make that necessary and how do we resolve them? The flaw is in the function of society itself, not necessarily the individual people. Money or lack of it is what makes this necessary for some. How do we provide access to things for free as a means of limiting theft, while still making sure creators are paid for their content and the value of that content is respected? We can find ways to make sure creators get paid, it’s only a question of who is paying. Technology and AI expose this problem because they make putting obstacles in the way more challenging.

Why does somebody generate every single social media post using AI? When sheer numbers of posts become algorithmically more important than less frequent authentic posting, it means we’re serving the larger system of social media. Social media is “full” in some respects and that means that to be seen it requires an individual to become a content factory to hopefully float to the top.

Having a post go viral without needing to post 12 times a day 7 days a week, is like winning the social media lottery. It’s impossible to predict. Without a guarantee of that occurring, people naturally default to what they believe works, which is just sheer volume. Unless you can afford to pay somebody to follow you around with a camera all day and then post it for you, you’re probably not posting enough to float to the top on most platforms. Are there exceptions? Sure because different social media sites work differently. It can be slightly easier to be seen on a platform like LinkedIn if you have content that professionals want and it can also be highly challenging to be seen on Facebook where most pages simply fall out of sight immediately. By making AI produce content, you get the chance at visibility without taking on a second or third full time job. The system creates that through its own structure. The environment we have now is a direct result of the environment we came from.

We used to live in a pay to play world. Visibility was expensive and only a handful could be seen. You no longer have to pay to play. Technology levelled the playing field while also exposing the original exclusivity problem. Now if I’m a really good writer or musician that’s on par with or better than somebody who has a major contract, I can gain exposure through social media. I don’t have to pay thousands for vanity publishing, I can pay $20 for an AI subscription that can write social media content for me based on the content of the book or the lines of the song I give it. It doesn’t guarantee success but it lowers the cost of the ability to at least participate in the process in a meaningful way.

The original system was great for those who managed to find their way in, but it also left out a lot of people with dreams and good work of their own. Eventually technology brings the gates down and people bust their way onto the scene that would never have otherwise had the opportunity. But that’s not the fault of AI. Once again, that’s the old system creating a false barrier to success and visibility through money alone.

The amount of attention available is finite. There are still a limited number of people that want the work being produced. Lowered barriers to visibility don’t change the fact that visibility is still limited by the number of people available to see the content. That’s not system created, that’s just a function of life.

What we have gained though is choice. If I don’t want to buy from a big chain, I can find a locally owned shop. If I no longer want to support major record labels, I can support people that don’t have a record label. Choice allows the attention to move away from large corporations and back to the individual artist.

Visibility is what allows the attention to move and that’s where social media plays a distinct role. It makes clearer why AI created content has become a reality. People are vying for attention in a world of limited attention, most of them without multi-million dollar budgets. In my view, giving more attention to independent artists is a natural rebalancing of power that is long overdue. That’s the part that technology creates that the old system could never accomplish.

We put a lot of time into creating barriers to access and far less time exploring why we need those barriers in the first place. There are many discussions still to be had about the role of technology and AI in society. But one thing is clear: we need to move the conversation away from trying to prevent the technology from existing and towards changing the conditions that make people feel they need it to game the system.

Discuss...