intel The lessons will continue until morale improves: Intel creates AI to read student's emotional state

[BachChain]

Summary

 

Intel, in partnership with Classroom Technologies, has developed an AI system that is capable of analyzing a video stream of a person in order to determine their emotional state. This system, as part of the "Class" software suite, is designed to integrate with videoconferencing platforms like Zoom used for remote schooling.

 

Quotes

Quote

It can be used to classify students' body language and facial expressions whenever digital classes are held through the videoconferencing application. Citing teachers' own experiences following remote lessons taken during the COVID-19 pandemic

Quote

The software makes use of students' video streams, which it feeds into the AI engine alongside contextual, real-time information that allows it to classify students' understanding of the subject matter. Sinem Aslan, a research scientist at Intel who helped develop the technology, says that the main objective is to improve one-on-one teaching sessions by allowing the teacher to react in real-time to each student's state of mind

Quote

It's interesting to ponder the AI technology's model - and its accuracy - when the scientific community itself hasn't been able to reach a definite conclusion on translating external action toward internal states. Building houses on quicksand rarely works out.

 

My thoughts

WTF? Who thought this was okay?

 

Sources

https://www.tomshardware.com/news/intel-students-ai-controversy

https://www.protocol.com/enterprise/emotion-ai-school-intel-edutech

[Zodiark1593]

15 minutes ago, BachChain said:

 

My thoughts

WTF? Who thought this was okay?

 

 

I mean, it's not like you can really legislate what gets coded. Whatever you can imagine, will probably get coded at some point or another.

 

Of course, in this application, morale is pretty important, and poor morale makes learning excruciatingly difficult.

[IkeaGnome]

This could be good. One teacher trying to teach 30 kids and pay attention to their own lesson won't catch every glazed over eye or confused person. Used for what it's intended for, it could be really good for students with learning disabilities as well. Early detection anyone?

[machin3gir1]

This seems more like an invasion of privacy and a slight breach of morality than anything.  Students that want to learn will make a genuine attempt to do so and it also depends on the teacher if this is true or not. 

Back in my year 10 maths class, we had a teacher that knew his math, but wasn't ready to teach our class (half our class was fairly disruptive), so we barely learnt anything. With the same class in year 11, we had a much better teacher and it showed- results went up across the board.  

I really hope Intel collaborates with actual classrooms and teachers rather than with their own executives who haven't been in the education system for several decades. 

[Arika]

26 minutes ago, BachChain said:

WTF? Who thought this was okay?

why is it not? There is nothing wrong with the creation of it, there are legitimate uses for such an AI.

The main thing that matters is how it's going to be used and if the students, in this case, get made fully aware of it's use. It's not breaching any privacy it's looking at someone's face and body language and making a prediction, you can go into the street and do the exact same thing to passersby, or more accurately a trained psychologist could go out and do this.

[da na]

Gotta open the HWiNFO log for my students holup

Hmmm... Specimen 13 is rather gloomy. I wonder the cause. 

[Elijah Kamski]

43 minutes ago, BachChain said:

WTF? Who thought this was okay?

Hey there,

I'm Eli and I'm from your government here to help with your child's education and mental health state as the AI has determined that your child is happy whilst playing games, but not whilst doing mindless work at school.

Please sign here to help your child achieve a sad, lonely, work oriented future, where fun does not exist, only the best future for your child.

[Elijah Kamski]

Out of a more serious note, I mean hey, what kind of company wants to see their future workers sad?

If you can't make them happy, it'll make everybody else sad, so best to find out if they are happy or not, and if you can't change them the usual route, well, time to code a way to make them "happy".

Here's to the future!

[Avocado Diaboli]

There was another thread a few months ago about something like this being implemented at Canon's Chinese offices that only allows workers to enter a building if they were smiling:

 

I'll reiterate my statement from back then: As the owner of a resting bitch face, I don't look forward to stuff like this actually being implemented in any meaningful way. Besides that, given that many people are incapable of reading other people's emotional states, I wonder what metrics an AI could use to make valid judgments.

[Alex Atkin UK]

I'm certainly in two minds, as if someone at school had noticed my emotional state (or cared about it) I probably would have done a lot better.   I was in a lot of pain, especially holding a pen/pencil, but I never told anyone as I didn't know that wasn't normal.

 

Had someone caught on to the fact I was in pain, I might have been under a lot less stress.

[Moonzy]

5 hours ago, BachChain said:

Who thought this was okay?

I think it's okay...?

 

What's the difference between an AI doing it and a human doing it?

And if it helps teacher teach better, why not

[Lightwreather]

This seems perfectly fine...

So longa s it isn't collecting personal data, but I doubt this will.

 

My only concern is for the teachers, Hopefully they won't have an Aneurysm when they see how many kids are bored in class. /s

[Rauten]

Honestly, this scares the crap out of me, specially if it gets used on children.

 

Like, sure, it could be a great tool to help teachers detect and support students that are having trouble. I can see the potential in it.

But how reliable is this, really? I feel like it could produce a lot of false positives and negatives.

 

What truly scares me though, is the potential abuse. This feels like it's just one "if" clause away from being exploitable and all kinds of horrible specially, once again, if used on children. <shiver> Brrr... what Facebook could try to do with something like this... </shiver>

[LAwLz]

1 hour ago, Rauten said:

Honestly, this scares the crap out of me, specially if it gets used on children.

 

Like, sure, it could be a great tool to help teachers detect and support students that are having trouble. I can see the potential in it.

But how reliable is this, really? I feel like it could produce a lot of false positives and negatives.

 

What truly scares me though, is the potential abuse. This feels like it's just one "if" clause away from being exploitable and all kinds of horrible specially, once again, if used on children. <shiver> Brrr... what Facebook could try to do with something like this... </shiver>

I mean... If it is not accurate we will find out about it and hopefully improve it. 

I am not really sure what abuse people think this will be used for either.

[Guest]

11 hours ago, Arika S said:

The main thing that matters is how it's going to be used and if the students, in this case, get made fully aware of it's use.

In other words, the other important concern is on whether the students or parents of students even consent into getting their students' faces scanned.

 

And, on whether they agree with the privacy policies schools using that technology made, which, when done incorrectly, could result in people not actually knowing what they're signing up for in the first place.

[Guest]

1 hour ago, LAwLz said:

I am not really sure what abuse people think this will be used for either.

What if that AI used for scanning students' faces be used to affect their future careers, then, from wages to even getting a job? What if they gather data from those faces for governments to use against us?

 

That's what I think becomes a real concern.

 

Have the AI be used for a different purpose, like figuring out depression or anxiety in a professional therapist's context, it would have been very different.

 

I see potential in helping people like me get a more accurate diagnosis than what humans tried to do, and yet they can be used against us. If AI gets used responsibly, eventually we could save ourselves.

 

That is, if we really want to make sure people don't use tech to further authoritarian goals.

[Blademaster91]

I feel like this could be open to a lot of misuse, like others here mention a student could be bored in class, or a teacher may not be teaching their kids so they are interested in learning the subject.

And this brings up the question of privacy and consent of scanning faces.

[LAwLz]

37 minutes ago, Herrscher of Whatever said:

What if that AI used for scanning students' faces be used to affect their future careers, then, from wages to even getting a job?

Do we have any precedent in the US that shows that this might actually be a possible outcome? I mean, how do you even suggest this will happen, assuming it will happen?

Remember, just because something might be possible, does not mean people will want to implement it.

 

37 minutes ago, Herrscher of Whatever said:

What if they gather data from those faces for governments to use against us?

How?

 

 

 

24 minutes ago, Blademaster91 said:

I feel like this could be open to a lot of misuse, like others here mention a student could be bored in class, or a teacher may not be teaching their kids so they are interested in learning the subject.

How is detecting bored students "misuse"? That is kind of the point of this. To detect students that are not engaged in the class so that they might be able to get help. I fail to see how a teacher gauging how engaged students are is "misuse". Students being bored in class is a problem, yet here you present it like students being bored and not engaged is what we should strive for, and certainly not take steps to fix.

[Rauten]

2 minutes ago, LAwLz said:

How is detecting bored students "misuse"? That is kind of the point of this. To detect students that are not engaged in the class so that they might be able to get help. I fail to see how a teacher gauging how engaged students are is "misuse". Students being bored in class is a problem, yet here you present it like students being bored and not engaged is what we should strive for, and certainly not take steps to fix.

The software is algorithmically detecting boredom.

The hardware, on the other hand, is detecting everything the child is doing. Facial gestures, where they look and for how long, how they move their arms, etc etc etc.

That's why I said that "it is an 'if' clause away from being abused". The only thing that keeps this good is how the algorithm behaves.

[LAwLz]

8 minutes ago, Rauten said:

The software is algorithmically detecting boredom.

The hardware, on the other hand, is detecting everything the child is doing. Facial gestures, where they look and for how long, how they move their arms, etc etc etc.

That's why I said that "it is an 'if' clause away from being abused". The only thing that keeps this good is how the algorithm behaves.

Again, you are not explaining HOW you think this will/could be abused.

 

And no, there are more more things stopping this from being abused than just the algorithm itself. In order for this to be abused there actually has to exist someone to abuse it.

 

Let's say the algorithm for detecting boredom is broken and it detects everything the students are doing. Every facial expression they make, every arm movement they do etc. Everything gets recorded.

1) How do you think someone will take advantage of this? If they don't have any feasible way of exploiting this, then it's a non-issue.

2) Who will take advantage of this? If nobody is interested in abusing this technology then we don't have to worry.

3) Why would someone want to take advantage of this? If nobody actually gains anything from abusing this then why would anyone actually do it?

[Guest]

1 hour ago, LAwLz said:

Do we have any precedent in the US that shows that this might actually be a possible outcome? I mean, how do you even suggest this will happen, assuming it will happen?

Remember, just because something might be possible, does not mean people will want to implement it.

I usually don't use US examples, but really, I didn't argue that this is happening now. All I did is asking the question to point out this can be use potentially for abuse.

 

Of course not every person would want to implement such tech for abuse, but the very few people who controls a lot of this whole economy and the world, you know, like Bezos, already used AI to further only their own interests at the expense of lots of people with, or looking for, jobs. It's a reasonable assumption to think they would be using tech like this.

 

1 hour ago, LAwLz said:

How?

Like, for example (edit: and yes I am going to use one US example just for this post), the FBI contacting a face collecting database to gather needed data necessary to track down civil protestors, the same way they've done so with Facebook.

Edited April 19, 2022 by Herrscher of Whatever

[darknessblade]

Let this AI loose on a "Woke" Campus and it will turn into a same trainwreck, of a Hate-speech Yelling AI as Microsoft TAY

 

https://en.wikipedia.org/wiki/Tay_(bot)

[MageTank]

2 hours ago, LAwLz said:

I mean... If it is not accurate we will find out about it and hopefully improve it. 

I am not really sure what abuse people think this will be used for either.

Oh, I can definitely abuse this.

 

Step 1: Buy dumb smart glasses

Step 2: Watch a lot of Poker tournaments to teach AI how Poker players handle their hands

Step 3: Join a Poker tournament

Step 4: Use fancy AI to read body language of fellow Poker players

Step 5: Go all in on 2-9 off suit and lose anyways

 

As you can see, pretty easy to exploit the practical application of this software.

[WereCat]

15 minutes ago, MageTank said:

Oh, I can definitely abuse this.

 

Step 1: Buy dumb smart glasses

Step 2: Watch a lot of Poker tournaments to teach AI how Poker players handle their hands

Step 3: Join a Poker tournament

Step 4: Use fancy AI to read body language of fellow Poker players

Step 5: Go all in on 2-9 off suit and lose anyways

 

As you can see, pretty easy to exploit the practical application of this software.

I can also see this being used on casting for actors. 

You can set the AI to trigger at certain threshold for detection for specific emotion and if the actor can't fool it, he / she won't get the job. 

[IkeaGnome]

2 hours ago, LAwLz said:

Again, you are not explaining HOW you think this will/could be abused.

 

And no, there are more more things stopping this from being abused than just the algorithm itself. In order for this to be abused there actually has to exist someone to abuse it.

 

Let's say the algorithm for detecting boredom is broken and it detects everything the students are doing. Every facial expression they make, every arm movement they do etc. Everything gets recorded.

1) How do you think someone will take advantage of this? If they don't have any feasible way of exploiting this, then it's a non-issue.

2) Who will take advantage of this? If nobody is interested in abusing this technology then we don't have to worry.

3) Why would someone want to take advantage of this? If nobody actually gains anything from abusing this then why would anyone actually do it?

I think the phrases others are looking for here are:

”guilty until proven innocent”

”once proven innocent, since it could possibly be abused, still guilty”

”AI bad”

This one is going to have a lot of conspiracy theories in it. 
Personally, as long as it’s used at face value this is something we can’t put value on. Hell, if we had this when I was growing up, it could have gotten me help sooner and preemptively stopped suicide attempts through that help. If it keeps even 1 kid from going through that shit, then I’m all for it. 

Edited April 19, 2022 by IkeaGnome