“That infringement threatens not just significant monetary damages, but the potential for injunctive relief that would threaten Twitter’s ongoing business as currently operated.”
This feels like an extremely important part of this story that no one is even discussing. What ended up happening here?
“Mr. Zatko alleges that:
Twitter’s platform is built in significant part on the misappropriation and infringement of third party intellectual property”
https://twitter.com/sparklingruby/status/1620625924097179649?s=20&t=8yV-HY-qn0MdTaQQOI5iTQ
Massive article here.
How much of twitter was run by AI/ML ?
Perhaps most of it. It appears, once they fed the system parameters...it ran with them and appears to have been so compartmentalized no-one had total control.
No wonder Elon could fire so many and have the system still so operational.
I think this is the battle Musk is fighting.
I remember @Jack 'feigning surprise' over Trump's removal...now I wonder if he could have stopped it if he had wanted to.
This is a massively deep dive.
Artificial Intelligence on Twitter:
An inside look at how Twitter used big data and artificial intelligence to moderate content:
Twitter created a data science team that was focused on combating U.S. political misinformation. This new division was created after the 2016 election. The team was comprised of data scientists who frequently corresponded with Trust and Safety at Twitter as well as third party government agencies including The CDC to identify alleged political misinformation. Content moderation and Machine Learning at Twitter was directly influenced by third party government agencies, academic researchers, and Trust and Safety.
The Ruby Files reveals an inside look into how Twitters’ machine learning detected for political misinformation, and the word lists used in natural language processing.
....How did Twitter define political misinformation?
“Every tweet in this set was judged as likely political misinformation. Some tweets were banned and some were not. This dataset might show how Twitter viewed misinformation based on words and phrases we judged to be likely indicators of misinformation. For example, stolen election was something we looked for in detections. We aren’t searching for fair election tweets. Get out and vote type tweets.
We also searched for the movie 2000 Mules, which said that Democrats stole the election. Our opinion of this movie was neutral. We didn’t judge it, but what we’d found was that tweets which contained that term were frequently associated with misinformation. If you tweeted about that, stolen elections, and a few other things, our AI would notice....
https://rubymediagroup.com/twitter-artificial-intelligence/
I wonder if that explains why trump capitalizes random words at times. I wonder if it does something to the AI. I mean other than also comms and confirming deltas and stuff
This article probably warrants it's own post by someone who can break it down better than I can.
https://rubymediagroup.com/twitter-artificial-intelligence/
I've tried twice to fix the spacing above.
My repairs aren't showing up.
I apologize for the mess it is.
Great read. Thanks for posting that link.