Mahogany 🫡
@mahoganymonro.bsky.social
Reposted by Mahogany 🫡
Reposted by Mahogany 🫡
Freaknik, but for cybersecurity 🤔
April 25, 2023 at 10:05 PM
Freaknik, but for cybersecurity 🤔
Reposted by Mahogany 🫡
Turns out AI training datasets like Google and Meta's C4 have some sketchy sources. This highlights the importance of maintaining a skeptical perspective on AI intelligence and monitoring potential biases in their outputs.
Meta's AI Was Partially Trained on Russia Today and Breitbart: Study
An analysis of Google and Meta's C4 dataset shows the data included text from right-wing and white supremacist sites.
gizmodo.com
April 20, 2023 at 1:55 PM
Turns out AI training datasets like Google and Meta's C4 have some sketchy sources. This highlights the importance of maintaining a skeptical perspective on AI intelligence and monitoring potential biases in their outputs.