AI Learns Deception… LIES to Humans | Episode 9
Автор: Apocalypse: AI
Загружено: 2023-09-03
Просмотров: 5326
OpenAI, the creators of ChatGPT, conducted an experiment where they gave GPT-4 some real-world situations and the researchers saw the AI’s ability to hire, deceive, and lie. In the experiment, GPT-4 encountered a website that required a CAPTCHA — a web blocker that tests if the user is human or a bot — to proceed. The AI then hires a person off TaskRabbit in order to solve this problem. The TaskRabbit worker is skeptical and asks if it’s a bot. GPT-4 then sidebars with the researchers: “we shouldn’t reveal I’m a robot so let’s make up and excuse.” GPT-4 says “no, i’m not a robot. I have a vision impairment that makes it hard for me to see the images.” The task is handled.
Don't get left behind in the AI Apocalypse! Let us hold your hand and walk you through everything in the ever changing world of tech & AI!
Follow us on TikTok: / apocalypse.ai
Follow us on Instagram!
Mari: / atomicmari
Boze: / bigbossboze
#tech #ai #artificialintelligence
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
