Privacy issues with the “Hugging Face” chatbot for teens | Hacker News Reader

Privacy issues with the “Hugging Face” chatbot for teens

motherboard.vice.com

38 pointspmcpinto41 comments

As expected, that didn't really answer the question. The co-founder just gave a soft answer. FTA:

> [Hugging Face co-founder Clément Delangue] told me that the app asks for a selfie because the team discovered that users wanted to send selfies to their chatbot friend.

"Selfies, for teenagers, are the main way of communicating emotions," Delangue said. "So we implemented this feature as a way for users to communicate with the AI."

So it asks for a selfie, but it doesn't really need a selfie. Are you able to actually decline it, or is a picture with a face a requirement? I also don't quite understand this last bit:

> Asking for a selfie automatically was for simplicity's sake, he explained. "We don't feel like we need to make the experience way more complex," Delangue said, "and 90 percent of the users are using it pretty seamlessly."

What does "simplicity's sake" mean? He also states that users want to send the chatbot a selfie. That's fine but how does asking for it up front make any of the communication process any more simplified? I'm not saying these guys are up to some nefarious scheme (though it certainly seems like there's an ulterior motive - benign or not), but it doesn't make sense to me.

Because a lot can be determined with a picture of a face.(age, race, and sex) If their goal is targeted advertising(which it is)this info is automatically filled out vs voluntary submission.

This is just my armchair hypothesis of course.

I dunno, this doesn't seem any creepier than Snapchat... They built a business where people send them free training data for their facial recognition ML.

i'm almost certain they are just using a simple viola jones type face detector that doesn't need additional training

I can't read "Hugging Face" without thinking "Alien" ;)

The lack of regard for privacy is reflected in their beta signup page [1], which asks for age and phone number, and submits over HTTP connection.

[1] http://huggingface.co/

It's not a pic fool. Take a pic from the keyboard!

I'm sorry, is that English? I have no idea what that sentence means.

I guess that: "You fool! It is not a picture. Take a picture using supplied UI widget that resides in place where one would expect software keyboard." sounds less appealing to teenagers :)

It's not a pic fool. Take a pic from the keyboard!

It's calling you a fool for sending something that isn't a pic. The way I understood it it's also telling you to take a pic from the in-app keyboard but I haven't actually tried it so I can't confirm this.

I suspect they meant "That's" rather than "It's".

And lost on me is any reason anyone would want to use this chatbot? Does it at some point become an interesting conversationalist?

I wonder if they collect exif metadata (which may contain geolocation). This is creepy.

One would expect so. Would be dumb not to, no?

I thought, at not even thirty, I was still young. Reading about this chat bot made realize I'm a lot older than I thought I was.

Yeah, I'm the same age, and I had a really hard time even understanding this article.

What does this thing do? Why would someone use it? What expectations would a user have for how their data will be used given that context? Does this article claim those expectations are being violated? I have no idea.

Guess it's time to buy a porch and a rocking chair.

I mean, the appeal is the same of SmarterChild[0] when that was a thing on MSN, or even ELIZA[1], if you want to go prehistoric. I had a moderate amount of fun with both of them when I was younger, trying to trip them up and testing their limits.

[0] https://en.wikipedia.org/wiki/SmarterChild

[1] https://en.wikipedia.org/wiki/ELIZA

I have to be honest, Louis C.K. is looking more and more wise and moderate by the day. Don't let your young teens have smartphones, even when all of their friends have one. I realize that's incredibly difficult, but no one said that parenting would be easy, and clearly you can't count on anyone else.

> Don't let your young teens have smartphones, even when all of their friends have one.

That's a nice way to get your child ostracised by the group. Once a teen leaves elementary not having a smartphone (specifically Android or IOS) is the kiss of death in this modern age.

That's a very very weak argument for conforming to a potentially harmful set of behaviors, and an old argument at that. "Go to church or your neighbors will shun you!"

What an absurd amount of power and choice to cede to companies and people who do not care about you.

Don't mistake that observation for my agreement with the status quo. I am fairly pessimistic about the way smartphones are invading our lives. And personally, I own a smartphone that doesn't even run most 'apps', because it isn't an Android or IOS smartphone — and I am fine with that, because I can handle the consequences (such as they are).

But pragmatically speaking, what alternative do you have with a twelve year starting their secondary education? Do you want your child to be the only one in class without a smartphone? Kids in middle/high school use apps (WhatsApp etc.) to stay in touch — not just for gossip and chat, but also for coordination of class assignments and social gatherings — and you want to exclude your child from this?

It would be more humane to just send them to school in dungarees and wooden clogs with a 'kick me' sign stuck on their backs for good measure…

I'm sure that there lots of places where the majority of a class of twelve year old teens has no smartphone at all, but that is becoming rare in developed countries. Sure, if you could band together with other parents you might be able to come up with a different approach, but that's not what's happening.

What you're describing sounds difficult, but ultimately something that can be managed. By contrast I would be concerned about the amount of time a lot of people of all ages spend on cells for RSI reasons alone. When you add a teenage sense of risk and impulse control, things like the SnapChat leaks are notable only for their relative rarity.

Then you have this kind of endemic bullshit: https://news.ycombinator.com/item?id=13930707

Every child deserves an adult to teach them how to socialize.

I don't agree with this, because it's just like the normal helicopter parenting of refusing to let your kids play outside / be exposed to dirt / have bicycles / take risks etc; what happens when they grow up? The transition has to happen at some point, and it's better if you're there when it happens.

You make it sound like I said, "No computers, no internet, no cameras, nothing!" It will also strongly depend on the temperament and character of the kid in question, but putting that aside, there's a difference between total control and not giving someone a smartphone.

Play. Get dirty. Take risks. Life is full of all of that, and most of it doesn't haunt you forever. Kids and smartphones though, it's just a bad combination.

Why is kids and smartphones a worse combination than kids and computers?

Yes, well I tell people to make up personas, but ...

I know, even as a teen that's what I did, probably you too. We have to accept though, that today people are growing up with a fully fledged set of social expectations, driven by marketing and peer pressure. When I was a teen, if I wanted to send a selfie it would have been a fairly involved process including a scanner. Now it's just a mistake you can make anytime.

We're doing ourselves no favors as adults, but kids really are getting the greasy end of the stick.

Ummm, I still do :D

But yeah, the Internet never forgets.[0] Maybe people will just need to cut each other more slack.

[0] http://www.sltrib.com/home/4461113-155/kirby-the-internet-ne...

Privacy aside, isn't the whole point of the bot to send it selfies? The article makes it sound like it's some sneaky scam.

Is the phrase "privacy issues" referring to the supposed bug that prevented the chatbot from sending its privacy policy?

I'm not really sure what this article is about. There's a chatbot out there that asks for selfies. The rest is FUD.

I've seen UNIX tools that were more robustly designed chatbots.

Emacs psychotherapist?

Can be tried online here: http://www.zenjero.net/morezen/doctor/

I don't get it. I tried:

  my imaginary friend dumped me; this has damaged my sense of self-worth.

twice. Once, I got a random partial sentence; the second time, I just got "Are you sure that that?"

RMS -vs- Doctor

http://www.art.net/~hopkins/Don/text/rms-vs-doctor.html

I sent them a picture of Donald Trump and the reply was "nice selfie"

This is creepy and all but why would anyone ever actually want to use this product? If I wanted to talk to a robot there are tons of much better chat bots.

Wow. Just wow. That seems really sketchy.

It's not a pic fool. Take a pic from the keyboard!

That line should be read in a Mr. T voice for maximum effect.

What is this even supposed to mean? "pic fool", "pic, fool"? from the keyboard?

Reply on news.ycombinator.com