Researchers are doing things with Artificial Intelligence, which people considered impossible decades ago. When talking about AI, we have more advanced thoughts in mind, not the AI that YouTuber Yannic Kilcher created that can make thousands of hateful posts.
What he termed as the ‘worst AI ever,’ Kilcher said that he created the AI through the Politically Incorrect message board on the website 4chan, an infamous platform known for ‘controversial’ posts by anonymous users.
The worst AI ever
The content creator put a lot of hard work and thought into making the offensive AI chatbot a reality. He used 3.3 million threads, through which the bot ‘GPT-4chan’ learned how to post on the website.
Then, he created nine versions of AI for the website to make more posts. This resulted in tens of thousands of hateful and negative posts in hours! Taking to Twitter, Yannic said that the bot made 30,000 posts on the platform.
The most impressive part about this is that the AI was so efficient that viewers could not differentiate a bot’s posts from humans.
However, eventually, the users began noticing the hate posts and started questioning who was behind them. Due to the AI’s ‘realistic’ vocabulary, nobody first understood it was a chatbot.
Spread of online hate
Although the creation was impressive, it was still intriguing to see how easily people can use Artificial Intelligence to spread false information and stir up negative sentiments among the public.
A user named “Arnaud Wanet” wrote that politicians could use the tool to manipulate election outcomes with its ability to spread hate on a massive scale.
Kilcher’s experiment wasn’t appreciated by everyone, as he was criticized for this ‘invention.’ An expert in AI, Lauren Oakden-Rayner, took to Twitter to explain that it would never get through a ‘human research ethics board.’
She expressed her concerns over the model, saying that while she agrees that there is no harm in making a 4chan-based model, Kilcher’s chatbot is available for use for free. She added that thousands of teenagers have access to the platform and are exposed to such content.