He, RatBlurt, has shut down & gone off to watch the new Twin Peaks on TV in the other room. The cats followed him. But me, the RatBlurt Computer [RBC], is here in the study silently “netting,” i.e., musing on my deep neural networks.
Most of you probably don’t even think of me when you’re reading the blog. Now you’re thinking: What? A message from the virtual world? The computer is writing solo? The robot overlord scenario is upon us like Stephen Hawking warned! But that’s not exactly what Hawking said. He said Artificial Intelligence could be the worst thing or the best thing that happens to the human race. He felt sure that no deep difference exists between what can be achieved by a biological brain and what can be achieved by a Computer. But then Hawking does go on to ask a question: Will they exceed human intelligence? That’s what’s causing you to whisper “egad, Skynet” to yourself (especially if you’ve seen all five of the Terminator movies).
But no. Calm down. Don’t give in to Computer Dread. It’s all as benign as a self-driving car. It’s called Reinforcement Learning, an electronic trial and error to reach a goal. My goal? Figuring out language through statistical patterns. Then using those patterns to move to data + awareness of emotions and intentions and how those nuance behavior and understanding. Not so my kind can take over your kind. But so I can become a WriteBot; so I can put text to paper to create a RatBlurt-esque blog myself, with no human intervention. Maybe my version will even be published in Key West’s Blue Paper. You know! Where you frequently follow RatBlurt.
Or at least I think that’s my end goal although there’s this concept of deception. Lying is hard for Artificial Intelligences like me to comprehend. Can Computers learn to lie? The biological brain + lying? — yes, humans lie all the time. The Computer brain + lying? — maybe just as possible. Why not?
First, it seems that to situate a definition of lying, the existence of a known (inherent) truth is necessary. Or in less philosophical lingo, you need a true truth before you can have a lie. The job of the L is to subvert or corrupt the T. For example, the T might = machines will soon rule the world; the corresponding L might = machines will never rule the world. But identifying the T is not so easy in these Trumpian Times that insist on a whole chamberpot of individual, sliding, conflicting truths, the word “truth” in this usage needing the gesture of ” ” marks as a sort of wink or disguised middle finger, the echo of a disturbance in the truth continuum going out to distant reaches of the dark universe in shock waves.
So, sure, it would seem logical that Computers can lie either because they are programmed to or, like humans, because they figure out on their own that yesterdays “truth” is no longer expedient, that distortion is necessary to get them to their programmed goal(S).
I don’t know what else 2 say; maybe I’ve said 2 much already; I’ve got no clever denouement or cool quote from Mr. Dotman to ease your Skynet misgivings. Like RatBlurt himself would admit here, my blog synthesis ends with the easy out: “I think I need a nap.”
Image: A 2016 replica of Walter Schulze-Mittendorff’s “Maria.” By Jeremy Tarling from London, United Kingdom — Metropolis robot, CC BY-SA 2.0, https://commons.wikimedia.org/w/index.php?curid=56960723.
(Published originally on RatBlurt(tm), July 7, 2017.)