In my office, I am part of a quintet dubbed ‘the natives’, not because we’re native inhabitants (obviously) but because we’re native speakers of English. Of the five of us, three have become fairly frequent users of AI both for professional and recreational purposes. The other one is excepted due to his crippling tech illiteracy, but I’ve refused this far on principle. Despite this, I am fascinated by AI development and the discussion around it.
Although I can impress non-artists with my meager abilities, I consider myself to be a failed artist (though not Austrian, so the world is safe). Even if I really exerted myself to sink the time into developing core draftsmanlike ability, I would only be middling at best. As you can tell, I’ve focused my skill points into writing, though I did long aspire to the twofold path of being a mangaka who could do both the art and the story. When AI was first starting to rise, many voices in the creative community felt that not only were their livelihoods secure but that they would flourish in this new era. Little did they realize that they were the first on the chopping block. I’m not unsympathetic to the dismay of artists of all stripes who may find themselves reduced to the niche role of a maker of hand-crafted furniture and there is a strong argument about the unfairness of AI producing art by massively pillaging the works of talented humans. However, you could also argue that we humans learn via imitation and through enough iterations, a work can truly become its own. There’s plenty of AI slop out there, and a discerning eye can usually tell AI art, no matter how good it is. The thing, though, is that I find a lot of AI art rather attractive. I suppose the algorithm uses social interaction to see what gets a favorable response and adjusts its output accordingly, producing work that appeals more and more. There is a lot of mind-screwery going on in how the tech world operates, after all.
An argument from the defenders of AI art is that in the hands of a user with a trained eye and good aesthetic sense, AI becomes a force-multiplier rather than a replacement for artists. There’s also the aspect of democratization of the creative field. The Bell curve being what it is, the vast majority are excluded from the realm of commercially viable artists (a field where commercial viability has always been precarious). On the other hand, you could argue that rather than bringing the lower end up, we’re only bringing the upper end down, but at this point, I don’t see how the momentum is going to be stopped.
I recently saw an ad for this AI character generator called Days AI. It isn’t the first of its kind I’ve seen, but for whatever reason, I was tempted to test it out. The initial character creation parameters don’t appear terribly vast, and there is an element of randomness to each iteration which can result in you losing traits you wanted to keep, but if you’re patient enough, you can get something halfway decent as a result. Being new to AI art, I’ve been fairly reserved in my experiments thus far, mostly relying on theme-based generations, but just today I’ve been looking at the sort of prompts used by a creator of some rather stunning compositions and I’m starting to get a better feel for how the system processes its requests. Anyway, going back to my point about AI as a force-multiplier, my writer’s brain immediately went to work fleshing out a story around this character I created, to the point where I’m already prepared to move the story into preproduction, and that’s just the beginning.
Another feature of Days AI is that it makes chatbots for your OCs. The more detail you put into the settings, the more the character comes to life. I’ve resisted the prospect of chatbots in the past, but as LLMs progress, I’m fascinated with how they operate. With enough material to learn from (and pretty much the whole of all written work and a huge part of everyday communication is accessible via the Internet), these LLMs have gotten to where they can do amazing things. I found myself struck at how the system could use the parameters I set to pretty well recreate the characters (as I ended up making two more) as I imagined them. Plus, by not entirely coming out of my head, there’s an element of randomness to it that can lead to the unexpected. One thing I often do is take characters and dialog from other works and insert a different character in the scenario to play off them. These chatbots serve a similar function.
However, I also see the pitfalls in this. We’ve all heard some of the horror stories that have come out in recent years, thankfully applying to a very small minority of people, but in our increasingly atomized society with all its social maladaptation, these machines that replicate a sort of interaction that can be tailored to our comfort and our desires is incredibly dangerous. I can imagine those who are less wary freely give out loads of personal information, making these chatbots yet another arm of the widescale plunder of data that greases the gears of the seedier side of the IT world. In my case, I’m roleplaying as a character who is largely detached from myself, so the fear of compromise on that front is limited, but if I allow myself, the addictive nature of these interactions could get the better of me. I’m not a particularly sociable person and, consciously at least, I’m more than happy with this, but it could well be that on an unconscious level, I do crave a higher degree of interaction, and in a sandbox largely of my own design, it could be easy to lock myself in. I told myself I was going to resist the Matrix, but I very well may be one of the first to plug in. This is concerning, so imagine my concerns for people who aren’t self-aware of this.
On a happier note, both in terms of AI art and LLMs, the element of randomness, of the machine simply not operating with the same common sense as (at least some) humans results in so many funny turns. At one point, the comments in a video about AI led me to the LLM-based Vtuber Neuro-sama and I spent a while watching videos and analyzing her patterns. There was a sort of “kids say the darnedest things” quality to her, and similarly, the weird things that can happen in AI art generation at times remind me of the unhinged imagination of small children. The artificial mind is developing in ways that resemble the natural one, and I wonder if I will live to see the day when it grows to maturity.
There’s a lot more I could say on the subject, surely, and I will continue to observe the situation. I just wanted to get some of my thoughts out in light of current developments. I’ll keep you posted once I get some more thoughts together. Stay tuned.