If you lead a team building a facial recognition project, and you either elect not to run/look at experiments on skin color, or you ignore your findings on skin color, and decide to sell that product to people with the express and upfront intention of using it to target the use of government-sanctioned force against individuals, there's no such thing as an "honest oversight." You have actively made a decision to endanger lives. Hanlon's Razor (which I think you mean rather than Poe's Law) doesn't apply when you know you're giving violent people a force multiplier.

The technological implementation is almost certainly an honest mistake but the marketing and sales around it is not. We've known about the importance of lighting in photography for more than a century and the computer vision industry has been battling the issue for decades in manufacturing because metal is shiny. Kodak even used to manufacture different films to better represent non-white people and camera sensor manufacturers today have entire teams with a swarm of diverse models dedicated to accurate color reproduction of skin tones across races. It's been an ongoing issue for people of color ever since TV became a thing because those technologies were initially very poor at color reproduction as well, but that didn't stop newscasters from jumping face first into televised manhunts.

We've been around this block several times before and while the quarry may change from photo accuracy to ML driven facial recognition, the hunt does not. There's no excuse for selling technology to industries facing (or creating) life and death situations when the bugs are so obvious.

This is EXACTLY what people are trying to raise awareness of. This is implicit bias. It doesn't matter if people have "good intentions" or made "honest mistakes" if the tool is implicitly biased. This is why having a diverse team working on a project is important, because it will help call out these issues early rather than after they have been put in production.

It's also not purely a technical problem, if you feed the model only pictures of white and asian male college students then it's no surprise when you get a model that biases towards recognizing white and asian male college students (which is exactly how several prominent models were trained).

Biased algorithms/models are particularly dangerous because they tend to provide a veneer of objectivity (plausible deniability if you want to put a more cynical lens on it) that could frustrate attempts to hold users accountable.

Agreed. A huge problem I haven't been able to think of is the (already happening in masse) practice of targeting black people because the data says so. However over time police Dept invests data capture resources in "bad areas" - which may have a lot of black residents. I wonder how to separate the defense that some black areas are obviously bad and why would a dept put surveillance resources in a good neighborhood?

I personally feel it's wrong but that's one thing I've always got hung up on in building a critique.

Mathematician Cathy O'Neil's book "Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy" is a good introduction to the implicit biases in machine learning:

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

That book is incredibly misleading.

This has nothing to do with machine learning. It is a simple correlational situation.

If African Americans have, on average, poorer credit ratings, then correlational models will begin to equate race with poor credit ratings, which will impact their ability to get credit and hence feeding back on that mechanism.

...of course RACE isn't allowed to be factored into financial applications, so the applications will often use other data points, like zip code, that end up having a correlation to bad credit as well as race. ...often producing the same result.

The problem isn't with the models - it's with reality.

The author famously said "Math is Racist". It's hard to get over such stupidity.

I don't think you're making the case you think you're making.

It seems like they are making the case that if reality is biased then models of reality will retain that bias.

If so they seem to make the point well.

If the only information you have about a loan applicant is where they live, your decision will be 'biased' if the location of where someone lives is correlated with other factors (as opposed to, say, the fact they live on a flood plain means don't give them a loan).

In this context, saying "Math is Racist" is like saying "Physics hates Fat People" because gravity disproportionately affects heavier people. Accurately reporting what is happening is not biased, making decisions without considering [edit: or not making a decision because you didn't consider] the context is biased.

Maths is a tool (well, collection of tools), and the onus is first on the tool user to use it in a fair way. Yes it is important for educators and tool creators to be mindful of how these tools will be used in practice, but there is a big jump from that idea to "Math is Racist".

Isn’t this a similar argument that could be made against race and gender based affirmative action? I don’t understand how organizations like ACLU are critical of face recognition tech because it reenforces implicit bias that engineers have but then turns around and supports race and gender based affirmative action that similarly reenforces implicit bias where PoC (but not Asians for some reasons) and non-males are presumed to be disadvantaged purely due to their identity.

I'm not sure what argument you are referring to here (if it was one above).

I think these organisations are criticising the tool builders for creating tools that are easily misused (or are created with unreasonable limitations, like only being valid for university students at one university, but are sold as widely applicable).

Supporting affirmative action initiatives like you list is trying to address the biases that exist in reality. I think this is often a bit backward (not addressing the root cause) but it can be expensive (in time, effort, money, politics) to address the actual root cause so these programs aim to address the bias at the place in manifests.

This is a similar (dare I say pragmatic?) argument to "it would be cheaper and more effective to just give everyone a no-strings attached payment each month then to provide means-tested payments to those who need help".

Detrmining if these arguments are correct is a different thing altogether, and I have no idea if these programs are cheaper and more effective then dealing with the root problem, or if it's even possible to define and address the root problem in the first place!

The two things you contrast above are fundamentally different - one is criticising tools and tool builders, the other trying to address perceived biases in the world.

When you say "[non-white/non-males] are presumed to be disadvantaged", have you talked to or listened to black or female academics? I follow ~4 black academics on twitter, and each of them have contributed to the #BlackintheIvory topic. Their identity plays a huge role in how others treat them.

> but not Asians for some reasons

Asian people are distinct because so many of them have immigrated recently, and immigration requirements favor educated and well-off folks. That masks many issues because they should have better than average outcomes due to better than average education and skills.

That's why "racism" has been redefined. Because it makes it morally convenient in the quest to "undo" past injustice.

On a side note: welcome to the Twilight Zone.

> so the applications will often use other data points, like zip code, that end up having a correlation to bad credit as well as race. ...often producing the same result.

You realize this too is illegal right? The law doesn't say "you can't use race" - instead it says (paraphrased by the Brookings Institute): "Are people within a protected class being clearly treated differently than those of nonprotected classes, even after accounting for credit risk factors?"[1]

O'Neil points out that math is often used to obfuscate this (whether it be deliberately or not). This is a valid point, and one that people who think of math as a values neutral tool should consider.

I didn't love the book, but it's difficult to make the argument that she is stupid.

[1] https://www.brookings.edu/research/credit-denial-in-the-age-...

> As an extension of this photographs of individuals with darker skin required more lighting than photographs of individuals with darker skin.

This is plain physics, no? Things are darker or lighter depending on the amount of light they reflect.

> I know that people are quick to jump to condemn something as racist but sometimes there really are just honest mistakes.

There are far too many things at play here. In a fair and just society, this kind of issue (like the XBOX Kinetic issue) would be met with a "oops, forgot to account for individual light absorption variance". A fix would have been issue and that's that.

Now, the problem starts when you begin digging. Why was this problem not caught? Because QA teams didn't catch it. Why didn't they catch it? Because the team wasn't very diverse, so testing failed to catch the problem? Why wasn't the team diverse enough? ... and now I've entered a societal rabbit hole that's far too complex for this post.

> I have a hard time believing that anyone at any level of the AWS structure set out to produce a racist facial recognition

Yes, highly doubtful. No benefit, and major issues if caught. It is far more likely that the dataset itself was biased. Why was it biased? ... and there we go again.

> we should sometimes temper the often strong reaction produced when labeling something racist, and focus our efforts on identifying and solving the issue rather than trying to act punitive.

Agreed, in principle. In practice, this produces no results in a mostly racist society. Companies (and politicians) will listen to outrage, they won't listen to well articulated and well-reasoned comments.

>> As an extension of this photographs of individuals with darker skin required more lighting than photographs of individuals with darker skin.

>This is plain physics, no? Things are darker or lighter depending on the amount of light they reflect.

Not quite. I don't know all the technical terms, but apparently, photograph technology early on settled on some standards and made some design choices that made it easier to photograph white than black people. That set a long-term precedent and standard for how film should work that persisted for a while and even to the present.

>>>Until recently, due to a light-skin bias embedded in colour film stock emulsions and digital camera design, the rendering of non-Caucasian skin tones was highly deficient and required the development of compensatory practices and technology improvements to redress its shortcomings. Using the emblematic “Shirley” norm reference card as a central metaphor reflecting the changing state of race relations/aesthetics, this essay analytically traces the colour adjustment processes in the industries of visual representation and identifies some prototypical changes in the field...

https://www.cjc-online.ca/index.php/journal/article/view/219...

Lay article on the topic:

https://www.buzzfeednews.com/article/syreetamcfadden/teachin...

I think the main thing is that even if the technology were improved and it could be proven that the biases were so low as to be negligible or that it basically identifies people equally well, the opposition would be on the technology as a tool.

People want some privacy in public. A tech that can track or backtrack people's movements is kinda creepy in a few ways.