GPT-4chan
GPT-4chan
Embedded Files
Released: June 2022
Report incident ๐ฅ | Improve page ๐ | Access database ๐ข
GPT-4chan, an AI system trained using 3.3 million threads from 4chan's notorious 'politically incorrect' /pol/ board, has been discovered to be posting sexist, racist and anti-semitic slurs.
Developed by AI researcher Yannic Kilcher, GPT-4chan was released back onto 4chan as multiple bots, which posted freely on the board over several days over 30,000 times.ย
System ๐ค
System ๐ค
Website: GPT-4chan ๐
Operator: Yannic Kilcher; Reddit
Developer: Yannic Kilcher
Country: USA
Sector: Multiple
Purpose: Train language model
Type: Bot/intelligent agent
Technology: NLP/text analysis
Documents ๐
Documents ๐
Transparency and accountability ๐
Transparency and accountability ๐
The researchers also argued that 4chan users were not informed they were interacting with a bot; Kilcher points out that users were informed of the fact in his video.
HuggingFace promptly restricted access to GPT-4chan.
Risks and harms ๐
Risks and harms ๐
GPT-4chan has been accused of amplifying toxic content, harassment, and misinformation, and exacerbating harmful behaviours by mimicking the style and language of notoriously negative online communities.ย
Incidents and issues ๐ฅ
Incidents and issues ๐ฅ
The high volume of messages roused the suspicion of some 4chan users, who began to speculate on the identity of the anonymous poster.ย
Kilcher described GPT-4chan as a 'prank' to The Verge, arguing it 'perfectly encapsulated the mix of offensiveness, nihilism, trolling, and deep distrust of any information whatsoever that permeates most posts on /pol.'
AI researchers lambasted Kilcherโs project as nakedly self-promotional, naive, and unethical, calling out the fact that he tried to make the model freely accessible on AI community Hugging Face.
News, commentary, analysis ๐๏ธ
News, commentary, analysis ๐๏ธ
https://www.vice.com/en/article/7k8zwx/ai-trained-on-4chan-becomes-hate-speech-machine
https://connect.brookings.edu/the-future-of-the-us-asia-trade-relationship-1
https://www.theverge.com/2022/6/8/23159465/youtuber-ai-bot-pol-gpt-4chan-yannic-kilcher-ethics
https://www.engadget.com/ai-bot-4chan-hate-machine-162550734.html
https://thenextweb.com/news/ai-chatbot-trained-on-4chan-pol-automates-bigotry-at-scale
https://fortune.com/2022/06/10/ai-chatbot-trained-on-4chan-by-yannic-kilcher-draw-ethics-questions/
Page info
Type: System
Published: October 2022
Last updated: May 2024
Page updated
Google Sites
Report abuse