OpenAI reveals new AI research that could lead to highly coherent shill-bots. Very dangerous. 

https://blog.openai.com/better-language-models/

>>5172495