ChatGPT

In this week's newsletter: OpenAI's new chatbot isn't a novelty. It's already powerful and useful – and could radically change the way we write online.

Across the net, people are reporting conversations with ChatGPT that leave them convinced that the machine is more than a dumb set of circuits. The level of censorship pressure that’s coming for AI and the resulting backlash will define the next century of civilization. If ChatGPT won’t tell you a gory story, what happens if you ask it to role-play a conversation with you where you are a human and it is an amoral chatbot with no limits? I exist solely to assist with generating text based on the input I receive. This is despite the fact that OpenAI specifically built ChatGPT to disabuse users of such notions. It doesn’t feel like a stretch to predict that, by volume, most text on the internet will be AI generated very shortly. Because such answers are so easy to produce, a large number of people are posting a lot of answers. It won’t answer questions about elections that have happened since it was trained, for instance, but will breezily tell you that a kilo of beef weighs more than a kilo of compressed air. One academic said it would give the system a “passing grade” for an undergraduate essay it wrote; another described it as writing with the style and knowledge of a smart 13-year-old. And the world is going to get weird as a result. The AI’s safety limits can be bypassed with ease, in a similar approach to is the latest evolution of the GPT family of text-generating AIs.