In my next book, I wanted to explore how an AI might be developed. As so often happens in science, I thought it might occur by accident. This is from Book Two of the Springbok Chronicles (untitled). In my previous book, The Huralon Incident, we meet the Reapers: deadly war machines fighting for the protagonists. Earlier in this scene the Reaper had reported 61% readiness, effectively non-deployable in the field. Brian, its technician, has been struggling to get it working again. HIP is described earlier and Human Interaction Protocol.
With a gesture, Brian made copies of the machine’s OS and firmware. He called up a block of code he’d written during a night of despair and anger over the bullying Grazny directed at him. For days after, every time Brian examined the code he wondered where his own mind was at. He had intended to write a piggyback OS that would stabilize Thunder’s often-collapsing system links by enabling auto self-correction. But as he looked at it again now, it seemed too simple in execution, yet too complex conceptually. It was something like a fractal organization system, simple in structure, but producing complex results.
The code was totally out of bounds for use in an operational weapon, but Brian kept wondering how well it would work. Plus, he’d added about 3K more lines of possible responses to human queries, and that would make his only friend a better conversationalist. Thinking hard, he drummed his fingers on the table. Thunder was hosed beyond belief. This couldn’t make it worse, could it? He loaded the code, triggered an overwrite, and rebooted the Reaper. He restarted its HIP and said, “Hey Thunder, what’s happening?”
“Feelin’ good, Bruh. You?”
Brian laughed out loud. That was more like it. “Aren’t you going to give me an operational status?”
Brian felt his stomach jump into his throat. His code shouldn’t have changed anything that much. “Say again?”
“74% now, and climbing.”
“No, no,” Brian muttered. He brought up the system logs, and sighed with relief. Core weapons release and action protocols remained standard. Then again, the arrays were expanding, allowing new potential behavior. Thunder’s same firmware error codes, the ones that had plagued it forever, still popped up, but the Reaper had created new software bridges that bypassed those problems. More bridges were popping up, hundreds per second.
His finger hovered over the diagnostic holo’s “kill thread.” Pulling on the virtual cord would trigger an emergency shutdown.
“What have I done?”
Brian’s eyes locked onto the folded up cube beside him. In diagnostic mode, the Reaper should remain folded, passive, incapable of doing anything else.
And then the alien-made killing machine stood up.
Brian yanked on the kill thread.
It didn’t work.
So what about you folks? If an AI were possible, how would it come about?