In another interesting news around GPT-3, a college kid created an entirely fake blog under a fake name using the AI model. We have extensively seen in the past how GPT-3 is the new latest obsession in the ML community.
Liam Porr, a computer science student at the University of California, Berkeley did this as a fun experiment and said that the scariest part of it was that it is super easy.
The blogs by Porr reached the number-one spot on Hacker News with readers engaging on the post and some even hitting ‘Subscribe’. While some people noticed that the blog was AI-generated, it highly went unnoticed. It suggested that the content produced by GPT-3 could fool people into believing that a human wrote it.
Sign up for your weekly dose of what's up in emerging technology.
To create a fake blog with GPT-3, Porr submitted an application filling out a simple questionnaire about the intended use. He collaborated with a PhD student to run a script. As reported, it gave GPT-3 the headline and introduction for a blog post and had it spit out several completed versions. The post went viral with over 20,00 visitors, and only a few asked him if the text was AI-generated.
Some of the blog topics it wrote are — “Feeling unproductive? Maybe you should stop overthinking”, “Boldness and creativity trumps intelligence”, among others. While on a few occasions the headline did not work out, but as long as he stayed on the right topics, the process was easy.
GPT-3 has been a debated technology with even Sam Altman suggesting that GPT-3 has weaknesses and makes silly mistakes. It has been feared to replace jobs such as journalists, coders, lawyers, accountants and more in the coming future.