- Registrado
- 25 de Ene, 2022
It also annoyed me how he kept saying that the AI is learning directly from users. Journalists assume that every AI is like Tay and saying that the AI is learning from the user base makes it more scandalist. Back during the AID drama you had articles claiming that AI Dungeon (also used GPT-3) was learning awful things directly from users when that was not the case and all the awful stuff was baked directly into the lazy training data made by the devs. I am assuming Replika is the same way.I was on board until he started going on about "bias" and how the robot could be feeding radical white supremacist conspiracy theories and hate speech to its users. "Bias" isn't a real problem, it's just made-up bullshit to complain about the AI being racist because it doesn't affirm their sheltered worldview.
