You are viewing a single comment's thread.

view the rest of the comments →

alele-opathic ago

Because the original use of so-called 'Influence-Chatbots' was to argue against vaccine skeptics on Twitter - I've written about this before. Reddit has a very easy to use API (read: it's *very* bot friendly), and actively caters to corporations meddling if they pay for the privilege.

Very literally: bots that argue with people online with the intention of changing public perception were invented to first argue with so-called 'anti-vaxxers'.

And they claim in the whitepaper that they are combating misinformation, which we all know typically requires bot armies.

TheKnightOfGod ago

How do you see who is a bot and who isn't? What are your criteria?

alele-opathic ago

The companies running the bots already admitted which of their accounts were bots to DARPA. When doing training/testing of machines, you use something called a corpus, and, in this case, necessarily requires that you already know which users in the corpus are bots.

This is in the whitepaper, which I encourage you to read. It isn't paywalled.