We’ve seen how AI can be used to generate very convincing deepfake videos, but can AI be smart enough where it can generate blog posts that sounds and reads like a human wrote it? Could it be convincing enough where it might even be able to fool readers into thinking that it was not written by a computer?

Turns out that it can, which is what comp sci student Liam Porr from the University of California, Berkeley did when he utilized the GPT-3 AI to generate an entire blog written by the AI, which was then convincing enough where his posts managed to top Hacker News with some people even subscribing to it.

For those unfamiliar, the GPT-3 AI was developed by OpenAI and is designed to be a language AI model. There were concerns that it could be abused so while it is available to the public, those who are interested in using it or experimenting with it will need to sign up to be selected to take part in a private beta.

Porr did sign up but is still waiting on being accepted, so he found a workaround where he managed to get in touch with a PhD student who had access to it and was willing to collaborate with him on it. According to Porr, it seems that he understands how easy it is to abuse and misuse the technology to generate clickbait content, which could drive down the overall quality of online content.

Filed in Computers. Read more about . Source: technologyreview

Discover more from Ubergizmo

Subscribe now to keep reading and get access to the full archive.

Continue reading