AI That is Too Dangerous for Release
An #AI that was deemed too dangerous is now released into the world. Is in “in the wild?”. (See my novel Threshold, due out in early 2020).
Researchers concluded that “GPT-2” was so powerful that it could create synthetic propaganda. Just what we need at a time where the United States is more divided than any time since the Civil War. “Due to our concerns about malicious applications of the technology, we are not releasing the trained model,” wrote OpenAI.
GPT-2 can analyze a piece of text, then predict the words that will come next. This additional content is, for the most part, indistinguishable from those written by a human being. Essentially this is an AI that can act like humans and create massive disinformation, like Fusion GPS or Adam Schiff, as a powerful bot. Now that is scary.