Posted on February 15th, 2018

Perverse Instantiation

How Artificial Superintelligence may impact humanity - a dialogue

Artifical intelligence

Jamie stood there, shifting nervously on the spot with a look, discernibly confused.

“You’re saying we programmed the machines to protect human life?” he asked, furrowing his brow.

“Yes, that’s correct,” responds Sam, almost mechanically.

“So why the hell are they chopping down all the trees?” growls Jamie.

“Well…as I understand, the trees were, um... in the way of growing manufacturing industries and the production lines, pumping out more machines.”

“Do they not know we need those trees to survive?!”

Contemplating the situation Sam, hesitantly and almost at a whisper replies, “I suppose we hadn’t entirely considered such effects on humans, and so, understandably the programming lacked such specificity. There were conflicting incentives. Safeguarding humanity… it could be said, was not the machines’ primary motive”.

“Not specific enough?! We gave you months and months to research, build, test, iterate and teach these machines”.

“It wasn’t long enough. I told you the team needed more time!” Sam stresses defensively.

“More time...and more money! Money that we didn’t have! Not to mention that, had we paused, even momentarily, you know just as well as I that some other company would have sold their machines to the manufacturing companies before we did. We would have gone under.”

Sam is silent.

“So how do we stop this?” Jamie asks. “They are stoppable right?”

“I believe it may be too late for that Jamie. We’re talking about a superintelligent species. We left a ‘backdoor’ in the code that we can try hack into, but the machines are continuously learning, their code always changing as they engage, ever increasingly, in recursive self-improvement. That backdoor may no longer exist…”

“We have to give it a try. Sam, we have to. I don’t want to be responsible for the death of the human race”.