https://blog.openai.com/better-language-models/
I just read this blog post from Open AI, and the sample articles written by AI. I was like so im...
For further actions, you may consider blocking this person and/or reporting abuse
This is awesome! The machine's writing style feels very natural.
It'll be interesting to see how these get used in the future. Imagine opening Google Analytics and having a small essay written by AI with detailed accounts of what's driving traffic and recommendations.
At the moment I'm choosing not to be scared.
Yeah, imagine that you just have to sketch out an outline, AI writes a novel for you.
I'm not sure why it would be scary, advances like these might help people write better articles, or improve spelling/grammar checking applications.
Obviously it would be bad if somebody made a fake text like this and presented it as true, but that would be bad even if a human did it. One should always check sources and so on.
They mention the automation of creating fraudulent messages, but most scam spam is written poorly on purpose to weed out the kind of people that are unlikely to fall for the scam.
it’s important to notice that those results were cherry picked. You can see that most of the results were the best in N tries and selected by a human. This means that the real level of the model is lower than what is shown.