RACEScience is screwed - the refusal to accept the results of multiple studies and experiments because it doesn’t align with political beliefs of progressives.

submitted by ClassroomPast6178 from quillette.com

sorted by:
top
newinsightfulfun
you are viewing a single comment's thread.

view the rest of the comments →

[–]ClassroomPast6178[S] 13 insightful - 1 fun13 insightful - 0 fun14 insightful - 1 fun -  (4 children)

The interesting thing was the AI being able to identify a person’s self-reported race at a rate greater than by chance without being deliberately trained simply by scanning medical imagery. The issue was that instead of saying “Hmm, that’s an interesting finding!” the researchers apologised profusely.

[–]Alienhunter糞大名 3 insightful - 1 fun3 insightful - 0 fun4 insightful - 1 fun -  (3 children)

It shouldn't be surprising. AI essentially isn't "intelligent" it's just able to process massive amounts of data and find correlations at a rate that far outpaces what humans can do. And since it doesn't really have "intelligence" it's essentially tossing in everything until it finds what works. Very brute force process of figuring something out but given enough time and data it can start to work.

I bet you could train a human to do the same thing if you showed various medical scans of certain races. There are clear physical differences between different groups of humans and you can probably get pretty good at guessing what race someone is by an X-ray if you know what to look for.

The thing is, it's still a social construct because you're programming the computer, and telling the people participating in the study, to choose from a list of races. The thing is that list in the first place is basically a social construct itself. We can choose to divide races extremely broadly at the species level, or we can go down as narrow as we can go and start to divide people down not only by what continent their ancestors came from but what country and possibly even what tribe.

So the question is at what level of granularity do you want to define "race" and that's the real reason why it's a "social" concept.

It's a social concept linked to a biological cause, but a social concept none the less because there's no one clear biological factor involved and you need to at some point start to define your classifications along somewhat arbitrarily defined categories.

[–]Dragonerne 4 insightful - 1 fun4 insightful - 0 fun5 insightful - 1 fun -  (2 children)

The thing is, it's still a social construct because you're programming the computer, and telling the people participating in the study, to choose from a list of races.

Not true.

You can ask the AI to group people into races with zero supervision or human social construction and the AI will by "coincidence" create the same races that humans use. This is because race is biological.

To say otherwise is to be stupid

[–]Alienhunter糞大名 1 insightful - 1 fun1 insightful - 0 fun2 insightful - 1 fun -  (1 child)

Source? The computer is programmed by humans, it begs to reason that it's going to be influenced by the human instructions. It's not that you can't tell who is genetically related to who by appearance, that's not the argument. It's a simple apparent fact that offspring often resemble their parents and given enough time and separation populations will develop various common traits within their population.

Race is by its nature a spectrum of sorts, though I loath the term because it implies some linear criteria one can measure where race is far more complicated. But I digress, we can take simple skin color as the criteria. And we can say people have more or less melanin. At what point do we draw the line of separation? That is socially decided.

In the same way different cultures look at the color spectrum and group colors differently in their own languages. To say that the physical reality of color, as electromagnetic radiation, is somehow a social construct would be lunacy. But exactly what shades of a color you determine to be "blue" or "green" is a social construct. Humans don't think in terms of wavelengths of light but in associations with things.

Many cultures have a tendency to combine the colors of blue and green into a singular concept of a "natural" color. Most cultures didn't have a word for Orange until people started to import the fruit. Which is why they share the word. The orange isn't named after its color, rather the color was named after the fruit, we teach that there are usually 6 colors in kindergarten. There's actually an "infinite" number of colors, but how you want to precisely define the difference is pointless, and generally falls along what our own social conditioning tells us to sort them.

Another good example is animals. Rats and mice are clearly different, yet go to some countries and the word for both is the same. Hell even the number of continents on earth is somewhat of a "social construct" since how we define the term is important. Kids in countries in Asia learn there are 5, Westerners learn there are 6. In actuality they are both right and both wrong. There's as many continents as there are land masses. But the Americas and Afro-Eurasia are just as easily considered to be two giant continents, or perhaps one if you disregard the bearing straight but I'll digress.

A lot of or understanding of classifications is in fact due to how we define things. And a lot of the time even our understanding of purely natural phenomenon can become entangled with various social concepts and indeed, constructs.

So I propose a simple yet extremely difficult question. What is an objective criteria by which we can sort people into different racial groups? And why is that criteria the best criteria?

[–]Dragonerne 3 insightful - 1 fun3 insightful - 0 fun4 insightful - 1 fun -  (0 children)

Source?

Because you asked for a source, I wont read the rest of what you wrote, because this is basic AI 101.

https://en.wikipedia.org/wiki/Unsupervised_learning

But here is a source for you to study up on.