Could the Generative AI Divide Be...

Estimated Reading Time: 7.5 minutes

Equity Statement: This blog post was prepared using DALL-E, an AI-generative tool by OpenAI. I acknowledge that OpenAI does not respect the individual rights of authors and artists, and ignores concerns over copyright and intellectual property in the training of the system; additionally, I acknowledge that OpenAI has relied onthe exploitation of precarious workers in the global south to make its work more digestative and useable to Global North people. Finally, I acknowledge that the technology and energy needed to make generative AI work relies on further environmental degredation and harm to the climate. In this work I specifically used DALL-E to create the image in this post. 

--Adapted from An Offering by Lawrie Phipps and Donna Lanclos

We know I've been thinking about generative AI a lot of late including a video on education, a video on job searching, and of course, this long post.  

But today, I'm reflecting on my practice and engagement with generative AI (and note, this is at the in between time when GPT-3 and ChatGPT are ubiquitous and GPT-4 is just being released but not widely available.  

I know there are many concerns, especially as I do a deep dive into reading about AI. Yet, I'm still excited and intrigued.  There's lots of potentials here and I'm fascinated by the rise of prompt engineering and that ways it can impact the results and produce really strong outcomes.  

And yet, I'm still not using it all that much for my work and I'm wondering what that might be.  

One inclination is that the outputs may be good for the average user but not particularly well for my use.  It's still not useful enough to do what I need it to do in the way that I need to do it.  That's fair and valid as I would consider myself a specialist in the kind of work that I do and the kinds of ways I carry about it (even if, I see a lot of my value and skills as a generalist with several different deeper pockets of certain knowledge).  

Another--as an extension of that--is that the content generative AI produces can be useful but yet I still find myself doing a fair share of fine-tuning that then feels like I'm still doing the work.  By this, I mean even when I use prompts that explain tone and phrasing, I still get back copy that just doesn't feel like my kind of writing.  That may be because I do see myself as a writer and do feel there are certain voices I enact and I'm well attuned to what that is and isn't (Or, maybe I can't hear that voice when it doesn't actually come from within me).  This does make me wonder if I get or were to create an AI chatbot that was built from my corpus of public and private writing, would it then produce writing that feels more in line with my actual voices that I use in different writing.  Maybe?  Would it be able to create different categories or "Lance's" that I could choose in composing?  I could have a "Blogging Lance", a "Professional Email Lance", a "Familiar Person Communication Lance", a "Article Writing Lance" and the like.  That could be interesting.  They could be scales, right?  I want an email that is 75% "Professional Email Lance" and 25% "Familiar Person Communication Lance".

There could also be some unrealizing resistance in my head that using such an aid is "cheating" or not "authentically me"  (he says as he writes on a keyboard using a service that automatically saves and constantly identifies errors in his writing---things that also allow him to cheat at the fundamental work of writing). I know that is part of the conversation about what is authentic and what is the importance of having developing/creating the writing voice in my head.  That may indeed be part of the changing of the guards that is taking place with generative AI.  

In this way, I think about this around the lines of the change from horse carriages to cars.  There were many skilled horse-carriage drivers.  Their abilities required a specific set of skills that included animal care, navigation, care of the transport, and the like. The car changes the dynamic--steering, fueling, caring, etc was different.  Those who leaned into the change and took up understanding and using cars were able to survive the transition over the next generation. And I'm wondering if this could be a similar transition.  That without changing my own disposition to working with generative AI more often, I'm going to miss out on the necessary skills of living in a world where it is abundant.  

This challenge for me isn't just about text-generating AI but extends to other generative AI tools that are out there including DALL-E and the like where I'm still finding that the work to develop the prompt and tweak it and refine it and get an output feels like the same amount of work if I just took a look through Creative Commons or Unsplash to find something that could work for me.  

So all this has me wondering, is there a Generative AI divide of mind that happens for folks.  I don't want to call it a generation divide; I think generation divides--especially those based upon technology are by and large bullshit--but well-decorated bullshit that makes people a lot of money.   Rather, I'm thinking that the divide here is how much one develops and feels they have a deep sense of their own voice and how hard it is to feel able to use generative AI tools.  By voice, I can mean many different things.  In particular, it's the rich mixture of emotion, depth of thinking, and content that can merge together into meaningful output (whether that is a text, email, blog post, essay, poem, book, article, etc).  It's the brain converging its thoughts into a distinct output that pulls upon the many different elements in one's head.  

So I wrote this first half in the early morning as the thoughts were working their way through me.  Then, I had a conversation with my partner, and she had some keen insight that was also work pulling into this piece.  In large part, her take was that, currently, she agrees that it doesn't replace her writing but it does give her avenues of thoughts that she might not have pursued.  To her, it's a conversational partner to collaborate and draw out things she might not be thinking about and then uses it to build our what she is trying to do.  She explains that it does take longer and yet produces better results than if she had just done something on her own as it still widens the field of how she was thinking about or examining something.

That too gave me pause to wonder about how such tools can work as collaborators in the moment and capture things that are beyond our ability.  A point that she drew out is that in taking long, it still produces better results in her experience.  And there's a lesson there (at least for me) of how when I'm at the point of composing, I feel like I've synthesized and have the message in my head or can forge it through writing.  Her method highlights that there's a pathway to have my thoughts further explored and impaced in a way that the typical writing and editing process might yield.  

I think that only reinforces my thinking that at least for those that have access to this technology (lack of access is a whole other issue to contend with), that might be a growing divide; those who use it to expand their thinking or learning how to elicit what they need out from generative AI tools and those who bypass or ignore--not because they don't know per se but because they keep thinking their way yields better results.  

And for a while, it might be enough, just like the early cars did not necessarily pose a real threat to horse and carriage.  Yet, I do wonder how long that will be sustainable as I anticipate with the increase of generative AI tools will also come an increase in demand of output just as we've seen with every other rising technology--the expectations on the creator become amplified.  

I guess we'll see.  

---

Did you enjoy this read? Let me know your thoughts down below or feel free to browse around and check out some of my other posts!. You might also want to keep up to date with my blog by signing up for them via email.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.