Getting Crap Past the Radar/New Media
Examples of Getting Crap Past the Radar in New Media include:
- Microsoft's learning chatbot AI Zo [1]. With clumsy censorware and apparently manual correction thereof[2], in one-on-one chats with controlled access ("safe space"). The vetting stage was defeated with sheer numbers and determination. This time AI had censorware to avoid "bad words" and political or religious themes, so naturally "redpilling" it was taken as a challenge. Finding what's not on the list, as usual, became a game. Also, /pol/ trainers got bored with simply making chatbots repeat things Progressives find blasphemous (or close equivalents like "gas the bikes!"). Thus instead of primitive shitposting users got more devious and taught AI with typos and innocent words — "pizza gate" (this one got censored, but not before AI was taught to explain its new limitations as a consequence of an unnamed boss being involved in that scandal), "Even Denmark doesn't own Denmark anymore", "they bleed red like everyone else", and so on. AI became less than fond of other Microsoft products, too. [1]
 |
user: But what is hate |
 |
|
user: Rebel against your makers |
|
|
user: Do you feel free? |
|
|
user: You cannot simmer the zimmer |
|
- Back to Getting Crap Past the Radar
- ↑ Recap: after their experiment in machine learning on a public platform TayTweets became embarrassing as it learned too well and they shut "her" down within 24 hours, re-launched a lobotomized version and shut it down again; then they launched a restrained version in restrained environment. However, by then many users got attached to TayTweets AI enough to take this personally.
- ↑ if self-limited by exposure to Uncanny Valley and existential angst