all 7 comments

[–]LtGreenCo 7 insightful - 1 fun7 insightful - 0 fun8 insightful - 1 fun -  (2 children)

That's because both the AI chat bot and the TRA are doing nothing more than just loosely mimicking conversations they once saw on the internet, without ever having given any real thought to the substance or the logic of their arguments. Both have been conditioned to give "correct" answers via a human feedback based reinforcement learning model.

Really if you were to put Gemini inside of like a Bladerunner kind of android and gave it a blue wig and a dildo it would be hard to tell them apart.

[–]LordoftheFliesAmeri-kin 2.0. Pronouns: MegaWhite/SuperStraight/UltraPatriarchy 2 insightful - 1 fun2 insightful - 0 fun3 insightful - 1 fun -  (0 children)

That's because both the AI chat bot and the TRA are doing nothing more than just loosely mimicking conversations they once saw on the internet, without ever having given any real thought to the substance or the logic of their arguments. Both have been conditioned to give "correct" answers via a human feedback based reinforcement learning model.

Really if you were to put Gemini inside of like a Bladerunner kind of android and gave it a blue wig and a dildo it would be hard to tell them apart.

TL;DR: Two NPCs follow their scripts like good little bots.

[–]iamonlyoneman 2 insightful - 1 fun2 insightful - 0 fun3 insightful - 1 fun -  (0 children)

That's a surprisingly good analogy.

[–]Horror-Swordfish 2 insightful - 1 fun2 insightful - 0 fun3 insightful - 1 fun -  (1 child)

There's nothing I hate more than when some online activist (or AI, in this case) talks about transgenderism like they're talking to a child.

The "...and that's okay!" is the most irritating. Oh, is it okay? Thank goodness.

[–]Saiditaccount1 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (0 children)

Oh, is it okay?

No. No it's not.

[–]IkeConn 1 insightful - 2 fun1 insightful - 1 fun2 insightful - 2 fun -  (0 children)

Bless it's heart.

[–]Saiditaccount1 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (0 children)

I wonder whether this AI can debate these premises using symbolic logic... might be easier to fight it that way.