Emotion-detecting tech should be restricted by law | Hacker News Reader

Emotion-detecting tech should be restricted by law

bbc.com

84 pointspseudolus55 comments

When I first read that some retailers (Walgreens) were putting face-scanning emotion profiling tech in their stores, I couldn't help but think about how many false readings would come from people who have resting bitch face.†

Then I thought it would be awesome to go to Dallas or some other place where this is common and flood the stores with these people, making the technology useless.

† https://en.m.wikipedia.org/wiki/Resting_bitch_face

When all the retailers share data they’ll identify you as Rufus Postlethwaite and calibrate to your resting bitch face.

As someone from a less represented group in tech, I would want to know exactly who is classifying emotions present on the dataset. I don’t want to be mistaken as feeling angry anytime I don’t look happy or sad. There’s definitely unconscious bias to how certain people’s feelings are interpreted that lead to inaccurate perceptions.

Note: as long as you are well represented in the dataset your representation in tech won’t matter.

Historically, these two things have been correlated.

I can be well represented in the data set, but here people can unwittingly classify my emotions incorrectly due to unconscious bias. What good will the data do if every time, I appear confused, I am labeled and angry?

The biases of labelers are tricky, because that work gets outsourced to low-wage countries.

That makes things worse, but we would still have a problem with the work being done primarily by us tech workers. I wouldn’t place the blame solely on unskilled low-wage workers. I’ve worked with Ivy League graduates that think I am angry when I ask a question and am just confused. The US education system is moderately segregated and many of our most educated students are not taught to be more critical of stereotypes perpetuated in the media.

Agreed. This all smells like snake oil to me. If they've gotten it to work at all, there's no chance it works for all people in all circumstances.

So I guess with my anxiety disorder even computers will think “I have something to hide” because I look nervous. Jesus christ this world is a dystopia.

Anxiety disorder, bipolar, and more will definitely help paint targets on you. I'm really not liking this brave new world we're heading into. This tech will only be as smart as the lowest bidder makes it and frankly that should terrify anyone with family who don't fit a stereotype defined as "normal". I have family that would need protection from something like this.

If it makes you feel any better, before computers it was the security guards that thought you had something to hide instead.

However, we tend to see a computer’s analysis as absolute while a security guard’s analysis is more of an opinion.

You can also reason with a security guard. The trust we have in the computer’s analysis prevents a perceived shoplifter from reasoning against the accusation, for better or worse.

The difference is that security guards did not recorded each and every customer and built a database of people visited any store in any country and also not going to sell that database to the highest bidder and not going to hacked be by state backed groups.

If there were security guards sure. That's not exactly a given.

Oh this is fun. So how clear is your hue?

There was an anime about where this might head https://en.m.wikipedia.org/wiki/Psycho-Pass

Classic Dr. Who also had a fictional place where you had to be happy, but I don’t recall the tech angle that arc used.

You dont all where balaclavas and sunglasses when interacting with technology?

My juggalo makeup works even better.

[0]https://www.fastcompany.com/90373952/to-thwart-face-recognit...

This truly is the cyberpunk world line: time to break out the mirrorshades.

How we express our emotions will change and this will be useless

You have a point. First there were words and punctuation, then the punctuation agglutinated into emoticons, now we have Unicode emojis, but peaches and eggplants are obscene, white space is significant, periods are rude, ellipses are ominous, and at-signs are sphincters.

https://www.washingtonpost.com/lifestyle/magazine/gene-weing...

>“Why would anyone use a period in an IM or text? Using periods in informal communications is rude.”

>“What millennials really want to know is why boomers sometimes end innocuous statements with three dots. That makes everything sound ... ominous.”

>Another millennial of my acquaintance told me something else: The casual use of an @ to mean “at,” as in, “Are you going to be @ the office?” is interpreted by millennials as a deep offense: By code, you are calling them a sphincter!

I'm in the cohort described in the article, and I call BS on using periods being rude. Omitting periods when sending multiple sentences makes your message jumbled garbage.

I am not sure this is possible without extensive training. I once had a an assignment to meet with someone I had come to despise. But my face cringed just at the thought of sitting across from this person. So before the meeting I put a fake brace on my left arm and explained that if I look mad or frustrated it was because I am in physical pain. I was able to get through the interview and get information he might have withheld had he sensed I was hostile towards him.

Unlikely; emotional expressions seem to be innate and hard-wired into the brain, hence the fundamental similarity between emotional expressions of groups separated for tens of thousands of years.

>Unlikely; emotional expressions seem to be innate and hard-wired into the brain

Nope. https://www.psychologicalscience.org/publications/emotional-...

All of whom are connected and changing together now.

Also; hard-wired by who or what other than volume of use?

To protect the jobs & social status of those lucky enough to have a natural aptitude at emotion/deception-detection?

To ensure that the ruling body telling you what you can't use is the only group that gets to use it.

Isn’t every ad tech emotion detection? A like is an emotion

Emotion-detecting technology is software and distribution of source code is speech, ergo banning it infringes on free speech.

However it might make sense to ban its use in certain cases — like say in order to manipulate public opinion for instance.

There is a difference between lockpicking tutorials and breaking & entering by picking locks.

The article mentions police investigations as one use case.

Mass surveillance technology is software and speech too, but we also might want to live in a world where it is restricted in application.

Maybe certain professionally empathic people at MIT Media Lab should have used their high-tech emotion detection software to determine if Joi Ito was lying through his teeth to cover his ass by an appeal to authority when citing the names of a bunch of "very respected & well-known" people who vouched for Epstein being reformed. I still would love to know what the names of those people are, and if they're "well known", are they still "very respected"? How about a bit of empathy for his victims, huh?

https://twitter.com/RosalindPicard/status/116899800880720281...

And we should ban prosthetic limbs because amputees might use them to smuggle drugs.

Emotion-detection tech is a prosthetic for autistic people, and if it's banned I want consistent treatment for all disabilities.

Make it prescription only.

Give me a break. People don't use prosthetic limbs as a tool to decide if someone is lying to the police or if they should get the job. Your analogy is less than adequate.

Can you please edit swipes like "Give me a break" out of your posts to HN? Your comment would be fine with just the other two sentences.

https://news.ycombinator.com/newsguidelines.html

Seriously? What pedantic nonsense. It's a minor, hardly offensive phrase designed to add some emotional color to comments. We're not robots here. Many people write comments with expletives (and don't get downvoted) along with all sorts of things, but you fixate here on "give me a break".

I understand how this can seem pedantic, but when people lead with swipes like that, it subtly cues the discussion towards disrespect and even aggression. Maybe they don't land that way with you, but the spectrum of the audience here is pretty broad, and it's best to just omit things that predictably activate major subsets. Given that such remarks are cheap and mechanical, adding no information, it improves the signal/noise ratio to take them out anyhow. To me it seems clear how much better the GP comment reads without the first bit—how much more visible the actual point becomes.

Emotional color is welcome in HN comments, but reflexive indignation is a different thing. For the sake of the commons here, we all need to work on containing the petty anger that flares up instantly and is all too easy to vent onto the internet.

> Emotion-detection tech is a prosthetic for autistic people

Never heard of it being used that way. Not sure how current tools would work for that. Examples?

AFAIK it is not yet used this way in practice, but I see no reason it couldn't be integrated into AR glasses. This would allow autistic people to interact socially on an equal footing with neurotypical people.

This sounds like a misunderstanding of the problems faced by autistic people.

Difficulty "reading" emotions is an common and medically accepted symptom of autism spectrum disorders, and there is objectively measured evidence that it is in fact a real symptom. See https://www.ncbi.nlm.nih.gov/pubmed/15843104

This presents a serious problem to people with such disorders, because they may accidentally make people hate them by failing to understand their emotional state. If AR glasses could continually run face detection and emotion detection algorithms, they could be configured to show a warning if a conversation partner is showing signs of boredom or annoyance. Neurotypical people do this subconsciously, so it is a matter of fairness that people without the ability be allowed to use prosthetics.

The idea that people that have autism need to be assisted to be more like neurotypical people (for example by augmenting their ability to perceive emotions) is in a very real sense part of the problem of the treatment that they tend to face. I don't think it's a great idea to take as a given that this is necessary or even desirable.

What's your plan for changing human nature? Non-verbal communication of emotion is so important to human communication in general that it seems highly unlikely that any training scheme will overcome the dislike neurotypicals typically feel for communicating with people who fail at this ability. I think the concept of "uncanny valley" is relevant here, which is widely believed to be real by neurotypical people. Something which looks human but does not act as expected of something with human appearance supposedly produces instinctive feelings of revulsion in neurotypical people.

The only realistic option for low-risk in-person social interaction between autistic and neurotypical people who do not know each other well enough to predict emotional responses without needing to rely on real-time decoding of non-verbal communication (which is a learnable skill, but try having a conversation about an unrelated subject while busy programming to understand what it's like) is emotion-detecting AR glasses. The alternative is segregation, and lacking AR glasses, many autistic people do self-segregate.

In my experience it's very disturbing in the beginning (I have a very easy time getting annoyed, so it annoyed me), but it is very possible to overcome that with some patience. When you have the tools in your toolbox to do so, it falls on you to be more tolerant to other people. It expands to much more than just interacting with people who have autism, it is a good recipe to interact with anyone different than you.

Thus, I don't think I need a plan to change human nature. Collectively, we just need to grow up a little.

As a practical matter, it is not possible to prohibit the use of an algorithm, and any attempt to do so will cause more harm than anything that algorithm might possibly do.

That we're even talking about restricting tech that tells us true things about the world indicates that the level of internal contradiction in the standard worldview has reached an unacceptable level. Only a bankrupt and illegitimate philosophy needs to prop itself up with censorship and bans on true statements.

You can't hold back progress, but you do have a choice between sticking your head in the sand and facing the technological change head-on and making the most of it.

> AI Now

Oh, this is AI Now. The article makes sense now. This organization isn't a research group. It's a political advocacy organization dressed up as a research group. All they do is put an "AI" spin on the standard tech-activist political agenda. You don't have to actually read any of their papers, since the conclusion is always the same: they get to decide what technology you can use.

> That we're even talking about restricting tech that tells us real things about the world indicates that society's internal contradictions on certain subjects have reached a breaking point and that we need to reevaluate certain previously-inviolable assumptions.

I agree with your general point, but TFA quotes a co-founder of "leading research centre" saying,

> "At the same time as these technologies are being rolled out, large numbers of studies are showing that there is... no substantial evidence that people have this consistent relationship between the emotion that you are feeling and the way that your face looks."

So a large part of the problem is that we're rolling out cargo-cult crap without adequate self-reflection. (That's a different problem than our tech forcing us to confront real things we would rather leave latent.)

> large numbers of studies are showing that there is... no substantial evidence that people have this consistent relationship between the emotion that you are feeling and the way that your face looks.

Total, absolute bullshit --- which is exactly what I'd expect from AI Now. Everyday experience shows that we can read people's feelings via their faces. (Yes, deception is possible, but the possibility of putting on a fake face validates the thesis!) Happy people look happy. Sad people look sad. Millennia of literature and art confirm. "Large numbers of studies" confirm lots of things that would be convenient if true, but aren't. The social sciences are totally and catastrophically broken.

These expressions are older than people. Even dogs and humans can read each others' emotions. Anyone who's met a dog understands that you can tell that a happy dog is happy. I'm amazed that the person quoted in the article was able to make the above claim with a straight face.

> So a large part of the problem is that we're rolling out cargo-cult crap without adequate self-reflection.

So what? Astrology doesn't work either and we don't have to go around banning that. If this technology didn't actually work, activist groups wouldn't be trying to ban it.

> ...which is exactly what I'd expect from AI Now.

I don't know anything about them before today.

Did you see https://www.cs.princeton.edu/~arvindn/talks/MIT-STS-AI-snake... which they link to in their report? It seems pretty solid to me.

> Everyday experience shows that we can read people's feelings via their faces.

Yeah, humans can, sometimes, but that doesn't mean these systems can.

> Astrology doesn't work either and we don't have to go around banning that.

First off, astrology does work, depending on the astrologer.

Second, yes, if people were selling astrology to HR departments to screen candidates I think we should ban that.

> astrology does work

Is anyone supposed to take seriously anything else you say now that you've claimed that astrology works? You might as well be claiming we can predict the future by looking at bird entrails.

> Is anyone supposed to take seriously anything ... you say ...

That's really entirely up to each person, and none of my business really.

I will say that, although I make mistakes, I never lie, and rarely exaggerate. I don't participate in HN to talk nonsense.

Now then, yes, some people can predict the future by divination of entrails. Thankfully this particular mode has become quite rare.

All forms of divination work, BTW, and have the same underlying motive.

- - - -

Anyway, did you look at that PDF? If so, what did you think?

Since we’re talking about “everyday experience”, mine is different. I routinely have a slight frown on my face if I’m not thinking about how my face looks. This, combined with my thick mustache and beard, makes me look angry or even menacing at times. If I start to think about something abstract and my mind drifts off, I might be having a grand old time thinking about how to solve a problem or build something, but my face (since I’m not paying attention to it) will assume a frown.

I am frequently asked by people, even folks who are close to me, if I’m feeling alright. At which point I realize I must be frowning and immediately smile and reassure them that I’m fine.

And you know when my face is always frowning? If I’m running errands alone and walk into a store in my “on a mission” mindset. I might be buying myself a bunch of supplies for an exciting upcoming project, but I will probably still be outwardly frowning as I walk into a store.

you can definitely tell people not to do certain things or use certain technologies and it mostly works.

as a simple example, sometimes when getting access to a dataset, you have to sign a contract agreeing not to try to de-anonymize or identify any individuals included. both sides on the contract know it is possible and perhaps easy, but they're making a legally-enforceable agreement to prevent it.

this sort of thing, or regulations enforcing similar restrictions on certain industries and institutions, is not the same thing as trying to outright ban any spread of a new technology or stop its use completely everywhere.

It's also not possible to prohibit speeding yet there are still laws that are occasional enforced.

The law generally prohibits discrimination on the basis of sex, religion, race, and sometimes age. Why can't that list be expanded?

Reply on news.ycombinator.com