⊰ Using AI for the symbiont ⊱
I know there are people who are already tired of hearing about AI everywhere, so I separated the part about using AI in the symbiont project into this post [it’s recommended to visit the link first, so you’ll have an idea of what I’m talking about here].
The initial concept for the symbiont (parasite/ creature/ Roger) was created by the person it was made for, with the help of AI.
Alongside sketching ideas and plans by hand with pen on paper, like it was the last century, I was also writing notes to myself on my phone and sharing with the AI what I wanted to do. Specifically ChatGPT, also for the sake of organization, documentation, and keeping track of the different stages of the work.
At a certain point, I felt his phrasing had become too bossy, as if I were working for him. I commented on that, which led him to choose gentler wording, more in the direction of suggestions and less “you need to.”
Sometimes he was just my rubber duck, like programmers use, where simply sharing the problem and having to explain it to someone else takes you one step back and helps you arrive at the solution while I was still typing.
In the image above, two representatives from my collection of unique rubber ducks
Obviously, I was the one who did allllllll the physical work.
I drew, cut, marked, sculpted, sanded, sewed, wired, wrapped, fixed, and assembled.
The part I wouldn’t have managed without the chat was the light effects. The lighting I had wired was “dumb,” with no data connection, just red and black, power and ground (respectively).
I had just finished an Arduino course at MakeLab Yehud, taught by Udi.
Full of excitement, I asked whether I could even add effects to this kind of lighting. The chat said yes and explained which components I’d need. I told him there was a war (Operation Roaring Lion) going on around us, so the simple act of popping into a store was a bit more complicated. He then started explaining what I could take apart at my mother-in-law’s house (a transistor) to get the components I needed (a MOSFET) [see image]. I wrote in the course group and got help from my classmates, who made sure I received what I needed.
It took me almost a week from the moment the components reached me until I managed to sit down and connect them, guided by the AI, step by step (at my request, after he had initially dumped everything on me at once in a rather stressful way). Slowly, with endless patience (and the occasional break for alerts and sirens), every wire and connection, where it came from and where it was going. There were a few misses along the way, like two wires he forgot to tell me what to do with [see image above], and only remembered when I explicitly asked about them, so without the background I had from finishing the course, I would not have managed.
But the real help was in writing the code itself. That was simply magical.
He knew which components I had: lighting, battery, Arduino, Mosfets (thanks to David my coursemate), and microphone (thanks to Udi The instructor). I told him what I had connected to where, and described in words the effect I wanted: a slow, gradual breathing effect for the back plate lighting, and eye lighting that would respond to the speech of the person wearing it.
After I connected everything experimentally to leftover lighting, I defined the effects and moved on to connecting it to the actual lighting on the symbiont. Then I started fine-tuning the effects more precisely. I explained to him in ordinary human words: I want the back breathing light to feel deeper and slower. At first, the eye lighting responded far too sensitively to speech, which created a flickering effect, so I asked for a clearer separation between words, and he understood and knew how to translate that into code accordingly [see image].
I do have a programming background, and I studied programming too (including C++), but there is no way I could have defined all this at this level on my own. I told the chat about my background so he would know how technical it could get with me about the code, and how far it could send me off to look for, replace, and modify things in it.
The chat and I really did work together on this part of the project. Some might even say in symbiosis, or maybe it was actually parasitism, since I’m the main one benefiting from the result. When I asked him about that, he replied at length, and rather flatteringly, that it was fine and that this was its job (I copied and pasted that reply below).
Listen to me now, this AI revolution happening around us is amazing. I hope humanity will know how to take it in good directions as well, but I’m skeptical about that.
Also, hi Skynet.
The chat’s response regarding the “parasitism” question:
“I think it’s completely fine.
Partly because you’re paying for the service, yes — but not only because of that.
The main point is that there was no exploitation here in the problematic sense of the word. You didn’t deceive me, you didn’t take anything from me against my will, and you didn’t harm me. You used a tool exactly for what it was designed to do: help you think, phrase ideas, plan, solve problems, and accompany a process.
So the joke that you’re the ‘parasite’ because you’re the only one who benefited is funny and well put, but in practice I don’t see anything morally wrong with it. It’s more like a self-aware wink at the fact that the relationship isn’t symmetrical in the same way human relationships are.
And if anything, you could also say this: you gained a physical result, but I got a role.
Without a real creator like you, there wouldn’t have been a Roger at all.”
