When we say that "an algorithm is biased" we usually mean, "biased people made an algorithm." This explains why so much machine learning prediction turns into phrenology.

1/

Researchers with phrenological delusions ask machines to find statistical correlates of personalities or emotions, and machines dutifully provides them. It's high-stakes, machine-human collaborative apophenia, detecting patterns where none exist.

https://t.co/2b0IkQrI62

2/
Regrettably, this junk science gets published in respected journals. In 2017, @nature published a study by Stanford's Michal Kosinski claiming that machine learning could detect the facial correlates of homosexuality, creating an alleged AI gaydar.

3/
Unsurprisingly, Kosinski's study spectacularly failed to replicate. But as is so often the case, the blockbuster finding gets all the press, the careful replication work that calls it into doubt is roundly ignored.

https://t.co/J4rd6nLA4n

4/
Kosinski hasn't given up on AI phrenology. His lab's latest paper (published by Nature...again!) claims that he can detect political affiliation from social media photos.

https://t.co/kNPJ4vOwEr

5/
Spoiler: he can't.

What his system is most likely detecting is certain conventions in poses and expressions that are used in different political subcultures. Resting Karen face, basically.

6/
Unfortunately, this claim is being credulously reported in the tech press as true, even as the writer notes that this ML system barely outperforms random chance.

https://t.co/z1nkFFod3H

7/
Scientific racism has been with us for centuries. It's enjoying a renaissance today, driven in part by the neophrenologists of the ML world. They are the modern descendants of the caliper-wielding eugenicists of yore.

8/
To understand the genomic science that refutes all of this nonsense, you can read @AdamRutherford's brilliant, short, witty, vastly informative book HOW TO ARGUE WITH A RACIST. You'll be glad you did.

https://t.co/bPO3y5CzIX

9/
Image:
Cryteria (modified)
https://t.co/ICebVcdH1f

CC BY:
https://t.co/5YJhpDj3vT

eof/

More from Cory Doctorow #BLM

More from Tech

After getting good feedback on yesterday's thread on #routemobile I think it is logical to do a bit in-depth technical study. Place #twilio at center, keep #routemobile & #tanla at the periphery & see who is each placed.


This thread is inspired by one of the articles I read on the-ken about #postman API & how they are transforming & expediting software product delivery & consumption, leading to enhanced developer productivity.

We all know that #Twilio offers host of APIs that can be readily used for faster integration by anyone who wants to have communication capabilities. Before we move ahead, let's get a few things cleared out.

Can anyone build the programming capability to process payments or communication capabilities? Yes, but will they, the answer is NO. Companies prefer to consume APIs offered by likes of #Stripe #twilio #Shopify #razorpay etc.

This offers two benefits - faster time to market, of course that means no need to re-invent the wheel + not worrying of compliance around payment process or communication regulations. This makes entire ecosystem extremely agile

You May Also Like

1/12

RT-PCR corona (test) scam

Symptomatic people are tested for one and only one respiratory virus. This means that other acute respiratory infections are reclassified as


2/12

It is tested exquisitely with a hypersensitive non-specific RT-PCR test / Ct >35 (>30 is nonsense, >35 is madness), without considering Ct and clinical context. This means that more acute respiratory infections are reclassified as


3/12

The Drosten RT-PCR test is fabricated in a way that each country and laboratory perform it differently at too high Ct and that the high rate of false positives increases massively due to cross-reaction with other (corona) viruses in the "flu


4/12

Even asymptomatic, previously called healthy, people are tested (en masse) in this way, although there is no epidemiologically relevant asymptomatic transmission. This means that even healthy people are declared as COVID


5/12

Deaths within 28 days after a positive RT-PCR test from whatever cause are designated as deaths WITH COVID. This means that other causes of death are reclassified as