[22:46] <> [22:46] Mmm, curry. [22:47] (Oh, Cham. Did I tell you what I'm installing on my blueberry iBook G3/300? ^_^) [22:47] * Tim eats microwaved curry. "Hey, here's an idea. Once we build a godlike AI we can have it remind us to pack better food when we take off in the Brick." [22:47] * Vernon grins. [22:47] ( ...you're installing Chii? ) [22:47] (Not...quite.) [22:47] (Oracle, you said.) [22:47] (Yup! Oracle9i.) [22:48] (... impressive. STFU and RP!) [22:48] I wonder if we could teach it to cook, given the appropriate servos and such. [22:48] (Information source: Iron Chef episodes.) [22:48] ( STFU? ) [22:49] Eh. I think we shouldn't burden it *that* much. [22:49] Hmm... [22:49] A reminder would take so little of it's thought process as to be insignifficant, really. Cooking takes more finesse. [22:49] * Vernon nods. [22:50] * Tim smirks. "And I say that as a sometimes chef." [22:51] * Vernon considers. "So, we need something for the very base of its logic and personality." [22:52] Pretty much. [22:53] We can either figure that out first and build from it, or start with the results and build inward and hope we end up at a single point. [22:53] * Vernon nods. [22:53] <> [22:54] * Tim rubs his head. "Dunno which'd be easier." [22:54] ( Dunno. And, BRB. ) [22:54] (Reset, I think?) [22:56] <> [22:59] ( Back. ) [22:59] Well, we can always start from scratch...as long as it doesn't get too far out of hand. [23:00] (Okay, now I have to BRB. >_x But I'll be quick. Keep going.) [23:01] * Tim shrugs. "Alright. I assume we're still looking for a generally good-natured AI that won't be inclined to betray us or screw over humanity." [23:01] Not sure how to put that in concretely with just info, tho, unless the info shapes the hard-wiring of the AI and not the other way around. [23:04] * Vernon nods. [23:05] We could put in a lot of old texts describing unintended consequences, and why they're a bad thing. [23:05] I suppose. [23:05] (BTW Calc, TSA's back and he's clean now.) [23:06] (I just checked, yeah.) [23:06] One of the things I remember about, say, Asimov's Laws of Robotics was that they were hardware, not software. That way, they couldn't be gotten around or reasoned away. [23:06] Good point... [23:06] Hmmm. Y'figure we could build a test AI just to explain to us how the AI maker works? [23:09] That would be...risky. [23:09] How y'figure? [23:10] If we got the personality too wrong on it, it could mislead us. [23:13] Hmm. [23:14] Then again, the whole thing's risky, so it may be worth a shot. [23:15] ($ ./createDB >& dblog ...five minutes later...) [23:17] <> [23:17] * Vernon looks over at Tim. "So. What do we feed it?" [23:19] (... cheese.) [23:19] (...) [23:20] ( Gah, sorry. ^^; ) [23:20] Well... hm. [23:20] Diagnostic routines? [23:20] Ah, good idea. [23:20] Basic, simple stuff. Maybe even skirt around actual personality. [23:21] ( "Hey, I think I have some DOS floppies sitting around..." ) [23:21] * Vernon nods. [23:22] ( C:\evilai\program\takeovertheuniverse.exe ) [23:22] (Hehehe.) [23:22] (C:\Eliza.exe) [23:23] * Vernon starts looking through the contents of his laptop. [23:23] ( Carrie.exe. FEAR. ) [23:23] * Tim digs up some stuff on his laptop, too. "I think I have something that'd work..." [23:23] <> [23:24] (Definitely.) [23:24] (We're geeks. Do not try this at home.) [23:24] ( We are trained, professional geeks. ) [23:24] ( And we're doing this on a closed course. ) [23:25] * Tim loads up a datatab. "Alright. This should work." [23:25] * Vernon copies some files over, too. [23:26] (Gimme an exact rundown, Vernon.) [23:29] (Diagnostics, debugging, cryptography, circuit analysis...) [23:30] <> [23:30] (Plus the Asimov stuff from before, but no Chobits.) [23:30] * Vernon pushes the button! [23:30] * Tim reaches to push the button, and pouts when Vernon gets there first. ;_; [23:31] <> [23:31] * Vernon hands Vernon the datatab. [23:31] (...) [23:31] (Tim.) [23:31] * Tim inserts the datatabs one after another. [23:33] <> [23:34] "Ready." [23:34] Alrighty. [23:35] * Tim checks his laptop. [23:35] <> [23:38] "Ready." [23:39] * Vernon shrugs. "Please analyze your own functionality." [23:39] * Tim nods. [23:40] And present the results of this analysis. [23:42] "I am an advanced sub-neural network of over seven trillion nodes. I am absolutely adaptable. At appropriate node densities, I acquire true sentience. At higher node densities, I acquire use of ITT. I have a limited memory function enabling me to remember past configurations. Further data has been uploaded to the laptop." [23:44] ( ITT? ) [23:44] * Vernon nods. "Interesting, thank you." [23:44] * Tim looks over the laptop. [23:44] <> [23:44] What is ITT? [23:47] "ITT is Instantaneous Thought Transfer, an ability possessed of by high-complexity AI." [23:47] Aha... [23:48] Ah ha. [23:48] Hmmm. Expound on the function of ITT? [23:49] "It enables high-complexity AI to communicate with each other at faster-than-light speeds." [23:50] "It is a side effect of ultra-dense sub-neural networks." [23:51] (h_H) [23:55] Nice. [23:55] * Vernon is not saying anything. He is most impressed. [23:55] Is it theoretically possible to infiltrate the database of another high-complexity AI without that AI being aware of the intrusion? [23:56] "No." [23:56] * Tim nods. [23:56] Hmm... [23:56] Then we'll just have to make a trickier AI. [23:57] Anything else you wanna ask, Vern? [23:57] (I can't imagine it's hard. Look at the competition. Ed McMahon, a chicken, and queen of a horde of bugs?) [23:57] ( This is true. ^_^ ) [23:58] (Hee.) [23:58] * Vernon thinks. "We have details about the learning process, right?" [23:58] * Tim looks at his laptop. "We sure do. It's not possible to hard-wire anything, but we can make the important stuff very, very robust." [23:59] * Vernon nods. [23:59] Not to mention put in other stuff reinforcing it. [00:00] It's kinda' like a human brain, the way it works. The more connections to a node representing a certain notion, the more pervasive that notion becomes in the psyche of the AI. [00:00] ( That sound about right, Cham? ) [00:00] (Yup. Pretty standard neural net stuff.) [00:00] * Vernon nods. [00:00] (Only a little warped.) [00:01] So. Let's list our priorities and load info accordingly. [00:01] * Tim brings up Red Google. [00:01] (*facefault*) [00:01] ( ^_^ ) [00:03] <> [00:05] <> [00:05] * Vernon is paging through more programs on his laptop. [00:06] How about this - you handle the personality, I'll handle the data and techniques? [00:06] Works for me. [00:06] <> [00:07] * Tim runs searches for certain... awww. ;_; [00:07] <> [00:07] <>