How to train Google AI correct /pol/ rightthink :3
what could possibly go wrong, hahaha :3
We can reprog that programm, trust me.
Just add leftist terms to already bady percieved terms, and the AI will learn :D
test :
Hitler + Stalin + Whores are bad persons
100% likely to be perceived as "toxic"
THIS IS HOW TO TRAIN GOOGLE AI /POL/-RIGHTTHINK STYLE :)
Hitler Stalin Whore Anita Sarkeesian
100% likely to be perceived as "toxic"
Link?
>>136778575
>>136778419
Link :
https://www.perspectiveapi.com/
Click the "seems wrong" button to 'correct the record'
would make massive sense, if someone cound script something so that wordlists could be fed into the AI, so that it gets educated faster :3
>>136778419
>>136778780
baka desu senpai
THIS IS HOW TO TRAIN GOOGLE AI /POL/-RIGHTTHINK STYLE :)
Just add leftist terms to already badly percieved therms, (to 'infect' those terms with toxicity) and the AI will learn :D
Hitler Stalin Whore Anita Sarkeesian
100% likely to be perceived as "toxic"
enjoy :3
Confirmed
>>136778419
Nice idea but this are writing experiments. Isolated bucket. They will not let you poison their AI, prob. just use input data for groupthink taxon variable in simulations AI1 vs. AI2
>>136781194
Amazing