ML Failures

Bursting the AI hype bubble, one prick at a time. FML

[ML Failures @FailuresMl]

OpenAI's obsession with the "safety" of its fairly bog-standard research has been part of its own myth-making from the start. There are very real convos to be had about dual-purpose #ML use; these ain't it

[ML Failures @FailuresMl]

Absolutely worthwhile read on OpenAI that confirms many of the suspicions others in the ML community had when the lab was first established #MachineLearning #AI

[Karen Hao @_KarenHao]

I spent half a year digging into @OpenAI, the SF-based AI research lab, originally founded by @elonmusk. I started with a few simple questions: Who are they? What are their goals? How do they work? After nearly three dozen interviews, I found so much more. https://t.co/wheK6r3r8j

[ML Failures @FailuresMl]

What happened to the @element_ai London office?? #machinelarning #AI #aiforgood

[ML Failures @FailuresMl]

Dystopia Friday news dump it is. Facial recognition phrenology with shit accuracy, the best of all possible worlds #FacialRecognition #mlFail #Dystopia #surveillance

[Clare Garvie @ClareAngelyn]

So much wrong with Clearview AI (as covered in @nytimes and @BuzzFeed), but a few thoughts on this accuracy "stat"—that it has a "98.6% accuracy" compared to "the world's leading providers" 1/

[ML Failures @FailuresMl]

Palantir is comically evil and inept, but that shouldn't distract from the fact that they're still capable of doing real & serious harm #MachineLearning #mlFail #Dystopia

[Mijente 🐜🐜🐜 @ConMijente]

Today, Palantir's CEO gave an interview at Davos. He admitted they help hunt down undocumented people.

This is after *months* of denial, saying they help fight trafficking & terrorism, deflecting when asked about ICE.

Gig's up @PalantirTech #NoTechforICE
https://t.co/3RfMmJdxAA

[ML Failures @FailuresMl]

👀 odds on this being informative and accurate?? #MachineLearning

[YouTube @YouTube]

Dive into the world of artificial intelligence in the season premiere of #AgeOfAI hosted by @RobertDowneyJr. Now streaming free on @youtube: https://t.co/7F6O46cqg2

[ML Failures @FailuresMl]

The entire thread is such a distressing example of this scientific community failing to hold its members to account. Looking at you, #Springer

[Paul Mozur 孟建國 @paulmozur]

@suilee The implications for surveillance are dark. So are the collection methods to develop the science. One paper published by Chinese police in a Springer Nature journal, used face scans and blood samples taken from 612 Uighurs in the remote town of Tumxuk, Xinjiang.

[ML Failures @FailuresMl]

👇👇👇👇👇👇👍 #MachineLearning #MLFails

[Arvind Narayanan @random_walker]

Much of what’s being sold as "AI" today is snake oil. It does not and cannot work. In a talk at MIT yesterday, I described why this happening, how we can recognize flawed AI claims, and push back. Here are my annotated slides: https://t.co/iCpyFw5urN

[Dan Roy @roydanroy]

I remember reading this paper when it first showed up an arXiv. (Was called l'arXiv back then.) Phenomenal work. I immediately recognized its importance and began working on its application to selling ads.

[Abeba Birhane @Abebab]

individual traits (intelligence, personality, etc.) are not located in brains so your predictions won't add up to anything better than racist pseudoscience clothed in #machinelearning and #neuroimaging

[ML Failures @FailuresMl]

I initially read this as "ruin", no idea why

[Martin Ford @MFordFuture]

Inside Amazon’s plan for Alexa to run your entire life
#AI
@techreview https://t.co/Y14HaymQvN

[ML Failures @FailuresMl]

How convenient that all these ethical #AI companies also get into trading. Today's in-no-way-a-conflict-of-interest-with-our-mission-statement courtesy of @element_ai https://t.co/iHK2uo98N3

[ML Failures @FailuresMl]

Next, @Google seems not to have learned from the @DeepMindAI & NHS #healthdata mishandling debacle. Go big or go home, better to ask forgiveness than permission, etc. From @Edpilkington @guardian #HIPAA #ProjectNightingale

[adversarybot @adversarybot]

A whistleblower has revealed that #ProjectNightingale has secretly transferred the personal medical data of up to 50 million Americans from one of the largest US healthcare providers to Google.

https://t.co/5Iry3AfRBc

[ML Failures @FailuresMl]

Inaugural posting on this account brought to you by recent healthtech fuck-ups.
First up, all is not well in the mythical lands of @babylonhealth #MachineLearning #fails

[Dr Murphy (aka David Watkins) @DrMurphy11]

Anyone happen to know who gave @babylonhealth its $1 Billion 'Unicorn' valuation?

I've got some magic beans that I would like to sell to them...

"Babylon Health - An absolute shambles of a company"
https://t.co/U7YJ6pTrDL