Developing a Federal #AI Standards Engagement Plan🧵
The Former Director of Twitters Responsible Machine Learning META division worked at Accenture.
Let's review some highlights from the below doc:
nist.gov/system/files/d…
Let's review some highlights from the below doc:
nist.gov/system/files/d…
"We encourage the United States government to adopt a similar whole-of-government approach" 
This paragraph on adversial attacks with #NLP sounds familiar.
It almost sounds like.. exactly what I reported in The Ruby Files re how Twitter weaponized NLP.
Adversial NLP attacks can come from within the organization - the bad actor is *not* always external.
It almost sounds like.. exactly what I reported in The Ruby Files re how Twitter weaponized NLP.
Adversial NLP attacks can come from within the organization - the bad actor is *not* always external.
Machine Learning Warfare:
"The model misinterprets the content of the image and misclassifies it. An attacker can tailor the expected behavior of an algorithm to achieve a number of outcomes."
"The model misinterprets the content of the image and misclassifies it. An attacker can tailor the expected behavior of an algorithm to achieve a number of outcomes."
"Adversial #AI targets areas of the attack surface that have never previously had to be secured, the AI models themselves.
From now on, organizations need to include these in their security budgets- or risk them being exploited by attackers."
From now on, organizations need to include these in their security budgets- or risk them being exploited by attackers."
"Poisoned training data"
"We would support NIST developing a sandboxing scheme for AI"
"....test and pilot AI algorithms and tools..."
"....test and pilot AI algorithms and tools..."
Next, let's review some highlights in the academic research paper:
If you read the research paper carefully, you will notice a common pattern emerge.
Policy rec. in the name of AI fairness- and a rec. to change results of the model if the output does not align w/ their definition of "fairness."
This is not academic research. It is activism.
Policy rec. in the name of AI fairness- and a rec. to change results of the model if the output does not align w/ their definition of "fairness."
This is not academic research. It is activism.
Machine Learning- Communism 3.0
"While equal opportunity is enforceable."
"While equal opportunity is enforceable."
In The #RubyFiles data- another pattern emerges.
NIST was repeatedly flagged as an N-Gram in the government agency category.
Why was Twitter using Machine Learning to monitor for NIST mentions?
NIST was repeatedly flagged as an N-Gram in the government agency category.
Why was Twitter using Machine Learning to monitor for NIST mentions?
Do you understand how dangerous this is?
AI “ethicists” have completely circumvented elected officials for well over 5 years now to implement whatever aligns w/ their worldview.
All in the name of “AI fairness”
AI “ethicists” have completely circumvented elected officials for well over 5 years now to implement whatever aligns w/ their worldview.
All in the name of “AI fairness”
I want to be clear in my language on this. When I say they- I am referring to a small group of people in ML who have hijacked the tech for nefarious purposes. I am not referring to the entire industry. There are many other people doing great work and find this deplorable.
I believe in the power of AI/ML. I want America to win the AI war. But we won’t lead with nefarious actors infiltrating the government.
They are using ML as a weapon to deploy personal ethics in the name of “AI fairness.”
It is unethical, immoral, and in any other industry-… twitter.com/i/web/status/1…
They are using ML as a weapon to deploy personal ethics in the name of “AI fairness.”
It is unethical, immoral, and in any other industry-… twitter.com/i/web/status/1…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
