Intel Calls Its AI That Detects Student Emotions a Teaching Tool. Others Call It 'Morally Reprehensible' (protocol.com) 38
An anonymous reader shares a report: When college instructor Angela Dancey wants to decipher whether her first-year English students comprehend what she's trying to get across in class, their facial expressions and body language don't reveal much. "Even in an in-person class, students can be difficult to read. Typically, undergraduates don't communicate much through their faces, especially a lack of understanding," said Dancey, a senior lecturer at the University of Illinois Chicago. Dancey uses tried-and-true methods such as asking students to identify their "muddiest point" -- a concept or idea she said students still struggle with -- following a lecture or discussion. "I ask them to write it down, share it and we address it as a class for everyone's benefit," she said. But Intel and Classroom Technologies, which sells virtual school software called Class, think there might be a better way. The companies have partnered to integrate an AI-based technology developed by Intel with Class, which runs on top of Zoom. Intel claims its system can detect whether students are bored, distracted or confused by assessing their facial expressions and how they're interacting with educational content.
"We can give the teacher additional insights to allow them to better communicate," said Michael Chasen, co-founder and CEO of Classroom Technologies, who said teachers have had trouble engaging with students in virtual classroom environments throughout the pandemic. His company plans to test Intel's student engagement analytics technology, which captures images of students' faces with a computer camera and computer vision technology and combines it with contextual information about what a student is working on at that moment to assess a student's state of understanding. Intel hopes to transform the technology into a product it can distribute more broadly, said Sinem Aslan, a research scientist at Intel, who helped develop the technology. "We are trying to enable one-on-one tutoring at scale," said Aslan, adding that the system is intended to help teachers recognize when students need help and to inform how they might alter educational materials based on how students interact with the educational content. "High levels of boredom will lead [students to] completely zone out of educational content," said Aslan. But critics argue that it is not possible to accurately determine whether someone is feeling bored, confused, happy or sad based on their facial expressions or other external signals.
"We can give the teacher additional insights to allow them to better communicate," said Michael Chasen, co-founder and CEO of Classroom Technologies, who said teachers have had trouble engaging with students in virtual classroom environments throughout the pandemic. His company plans to test Intel's student engagement analytics technology, which captures images of students' faces with a computer camera and computer vision technology and combines it with contextual information about what a student is working on at that moment to assess a student's state of understanding. Intel hopes to transform the technology into a product it can distribute more broadly, said Sinem Aslan, a research scientist at Intel, who helped develop the technology. "We are trying to enable one-on-one tutoring at scale," said Aslan, adding that the system is intended to help teachers recognize when students need help and to inform how they might alter educational materials based on how students interact with the educational content. "High levels of boredom will lead [students to] completely zone out of educational content," said Aslan. But critics argue that it is not possible to accurately determine whether someone is feeling bored, confused, happy or sad based on their facial expressions or other external signals.
AI for Autists (Score:5, Interesting)
Tell me, if you can't tell whether someone is bored or engaged by looking at their face, how are you going to label data to train a model to do the same thing?
This is pure nonsense. Maybe it has some edge case for people with autism that really can't read other people's emotions.
Re: (Score:2)
Tell me, if you can't tell whether someone is bored or engaged by looking at their face, how are you going to label data to train a model to do the same thing?
This is pure nonsense. Maybe it has some edge case for people with autism that really can't read other people's emotions.
Except your first point still holds. If it is impossible to train the AI then it will not work. For your second use case that makes it worse than useless. It will actively hinder understanding.
Re: (Score:1)
Yeah, there are two cases
1) You can label the faces - and train a model (useful for autists, but no one else)
2) You can't label the faces - useless model
Re: (Score:1)
Re: (Score:1)
Re: (Score:1)
Garbage in, garbage out.
Re:AI for Autists (Score:4, Interesting)
I suspect it comes down to something like. I can sit there and keep an eye on teacher and make sure to sit up strait an smile when s/he turns from the black board to face the class. However while s/he is up there writing or talking to some other student I don't have to be acting. The computer on the other hand can watch me all the time...
Its kinda like those keyboard monitors and screenshot grabber surveillance applications awful companies subject their workers to. Nobody wants to live in a anticipation like that. I dont know maybe in grade schools something like this would not be matter to much with per-adolecents. However I can't think of anything that would have functioned as a more powerful incentive to make me consider dropping out in high school than some mechanized system constantly trying to detect personal feelings I did not want to share.
Re: (Score:3)
There is a test we can use to see if it is useful. The test is to implement it throughout Intel including, and especially, in the executive offices. I think it would make sense to watch them continuously to see if they are engaged and not pursuing "other" interests. And naturally, it must be on all the time for an effective test. The results get published to the stockholders.
AI for Online Teaching (Score:2)
Tell me, if you can't tell whether someone is bored or engaged by looking at their face
What if you can't see their faces either at all or at least well enough to tell, for example when teaching an online course? I agree it would be useless when teaching in-person. However, I could potentially see some use for this in online courses where it is extremely hard, if not impossible, to get the immediate feedback that you get when lecturing in-person.
Re: (Score:2)
There may be subtile markers that humans can't detect but a generically trained AI can.
Re: (Score:2)
You think an "AI" can outdo millennia of evolution? You should do something about your unfounded belief in tech.
Re: (Score:2)
You think an "AI" can outdo millennia of evolution? You should do something about your unfounded belief in tech.
It's already doing that? For example person recognition from their gait? How's your "billion years of evolution" doing with that? Because for AI that's piece of cake.
Re: (Score:2)
You misunderstand the problem. Hence your statement is complete BS.
I'm guessing it doesn't work on minorities either (Score:3)
Mind reading by facial features and body language does work... for salesmen. That's because they've learned how to read the body language for a very specific purpose and with a very specific and very local group of people. I don't see it scaling up to a University setting where so many different types of people
Re: (Score:2)
Exactly. Also, salesmen may need quite a while to "lock on" and have a pretty high failure rate in general.
Re: (Score:2)
Tell me, if you can't tell whether someone is bored or engaged by looking at their face, how are you going to label data to train a model to do the same thing?
By using something other than their face.
For instance, you could ask them.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Indeed. Pseudoscience that just took a problem and then lies about being able to address it. I find out what my students understand in the exercises. The students themselves do too.
Also, this is obviously not about teaching, that is just the lie used to sell it now. This is about interrogation support software (i.e. "lie detection"), ad targeting and manipulation of voters.
Re: (Score:2)
Tell me, if you can't tell whether someone is bored or engaged by looking at their face, how are you going to label data to train a model to do the same thing?
This is pure nonsense. Maybe it has some edge case for people with autism that really can't read other people's emotions.
FFS. Make some photos of people during class, then have them fill out a questionnaire after on how engaged they were. Here's your training data. Later you can track the engagement level of class on real time, without having to run questionnaires every 5 minutes.
7th Grade - College Undergrad Sr year. (Score:4, Funny)
Emotions:
Horny
having a crush on someone.
deliriously happy because the crush likes them
depressed because the crush doesn't like them
distracted by hobby or interest from being horny
board because the class is wasting their time from their crush
angry at person seen a competition to your crushes affection
planning something to get your crushes attention
pretending the don't have these feelings, because everyone tells them that they are wrong.
Re: (Score:2)
assignment: bored vs board (Score:1)
Compare and contrast the definitions for the words "bored" and "board".
Use each word correctly in at least three sentences.
No exemptions! The word "bored" is used in the summary.
Re: (Score:2)
When I am bored if like to hit grammar Nazi with a board.
The hole in your head is bored out from a sharpen board.
Perhaps if I am bored enough, ill fix Firefox default spell checker, perhaps the problem is listed on their scrum board.
Oxymorons'R'Us (Score:5, Funny)
Re: (Score:2)
Re: (Score:2)
Well, according to some people that know, educators are among the most stupid, disconnected and arrogant people on the planet. Hence they may be an ideal target for this scam. Caveat: I have quite a bit experience as a part-time academic educator.
Next let's point a camera (Score:2)
at the teacher coupled with an AI that detects his happiness level when little Timmy erects.
As an added benefit (Score:2)
tells if someone is a witch.
Creepy factor is high on this one. Glad it was never used on me.
Emotion Recognition DOES Work, But... (Score:5, Interesting)
Emotion recognition does work, but not in arbitrary contexts. A classroom may be one of the easier environments for such a tool, as everyone is supposedly engaged in the same task at the same time.
I have a friend who has worked in this exact field for well over a decade, initially as a mathematical theorist, then to working with a team to apply the theory. She moved from academia to a well-funded startup, which was then bought by a Fortune 50 tech company (not Intel).
She had lots of stories concerning the immense issues of actually getting good data for the system. Humans, and human emotions, are WAY too complex to start with. So they searched the animal kingdom for simpler surrogates. There were two animals that soon rose to the top: Cows and horses.
Yup. Who knew? Both cows and horses have straightforward emotional systems with easily detected facial expressions. Much initial progress was made, to the point that the startup received investments from both animal rights groups ("Are your animals happy?") and animal husbandry companies ("Are your animals stressed, leading to lower production?").
This animal progress triggered renewed progress on recognizing human emotions. Evolution has conserved surprising amounts of commonality of emotional facial expression among the mammalian branches. It allowed "hard-wired" emotions in humans to be detangled from more culturally related expressions of "feelings".
It is also the case that determining a distribution of emotions across a crowd is easier than for any single individual. An instructor watching the shift in emotional sentiment across a class as a whole could be very useful!
Which begs the question: To what extent are classrooms like stockyards?
Technically Intel is correct (Score:1)
Yes, it's a "teaching tool" to train their neural net bots.
Free built in AI (Score:4, Informative)
The hard part is finding the root cause and adjust your teaching methods.
Wear a face covering. (Score:2)
Nope. Wear a face covering. Maybe don't don it until the lesson starts, so they can at least be sure you are really you. But the moment they want to track engagement? Face covering on. This invasive tech should not be tolerated. Teachers who are unwilling to ask their students questions to determine if they understand, or to hold quizes, aren't doing the job right.
The for profit surveillance state (Score:2)
Re (Score:1)
AI in educational process (Score:1)