Skip to main content

I replied to all my Tweets in my timeline using ChatGPT.

By writing a connector between ChatGPT and Twitter, I was able to reply to every tweet in my timeline with only bot responses. With the help of the Selenium web driver, I was able to control a web browser and interact with ChatGPT without using any API.

Interestingly, the bot replied in different languages and often created valuable content in the conversation. Like the following one:

This caught the attention of some major accounts, which retweeted my tweets and spread them to a much bigger audience:

Analytics after retweets.

However, there were some drawbacks. My online community was able to spot that the replies were not written by a human:

ChatGPT is a game changing technology and soon OpenAI will change its pricing model from free to a pay-per-API-call system, similar to how GPT3 works.

This made me think: If we are able to pass the Social Media Touring Test, are we going to be able to trust content in social media anymore? Solutions that spot bots by checking the type of content may not be enough, and the only solution I can think of is to use techniques like captchas.

Elon Musk has his work cut out for him – will we have to solve a captcha for every tweet we write in the future?


Comments powered by Disqus