[ForHackernews]

I mean, it's an EU-funded academic site designed to raise awareness of privacy issues, so I'm inclined to believe them when they say they aren't keeping any of this data.

Maybe it's because programmers tend to think categorically, but I find HN frequently has this problem where people are freaking out like "why is it okay when Mozilla does X but you don't want Google to do X?!" as though we're supposed to treat every question as some sort of sui generis scenario divorced from all history and context.

History matters, context matters. I (generally) trust academics and democratically-elected governments, and I (generally) distrust giant ad-tech companies and authoritarian states.

[unityByFreedom]

Context mattering has nothing to do with sharing your face with a random site that claims to be supported by "The EU".

This is a site made by one guy, not the EU.

[socialcooling]

Creator here.

The project was created as part of my role as artist within the Sherpa consortium. It's a Horizon2020 research project whose goal is to figure out what Europeans believe are the biggest issues around AI that we'll experience in 2025. Most of the other partners are universities, and I'm the lucky one who gets to translate what we learn into art pieces.

Learn more about Sherpa here: https://www.project-sherpa.eu/

All the other works I've created / am creating for Sherpa can be found here: https://www.sherpapieces.eu/

Besides working as an artist I also work as a privacy designer, which means I specialise in creating things/products that protect privacy.

This website was a very cool challenge to build, a testament to the power of javascript.

[adaml_623]

I followed a link to socialcooling.com as well the other day. The things you are talking about are critical for a decent civilisation moving forward. Thank you for talking about them and making us think.

[ForHackernews]

This is really cool work! If somebody else wanted to get involved in working on projects like these, how would you suggest getting started?

[socialcooling]

That's difficult to say. You could start with getting to to know an issue and specialising in it. Reading books from academic writers (not the stuff sold at airport bookstores). Read Slashdot. Once you've got a grasp, think about how you could translate what you learnt so that your mom would understand it. Think about what your mom likes: quizes, human-interest stories. Lighthearted stuff. Then create your own translation.

It's probably easiest to connect with a local group of people who care about these issues.

[drpgq]

Where do the algorithms come from?

[socialcooling]

The main one is FaceApiJS.

One I trained myself (BMI), and the others I just scavenged from existing Github projects. So with most of these I don't know how they were trained, what photos they were fed, etc.

[drpgq]

How accurate is the BMI estimator? I trained one years ago, but the performance wasn't great.

[ForHackernews]

Then I guess it says something negative about today's internet that I trust some random guy a lot more than I trust Facebook or Google and their hordes of lawyers. If he says he's not collecting my data, I'm inclined to believe him. If Google says they won't misuse my data, I assume that comes with 200 pages of legalese loopholes to let them do whatever they want.

I don't think I'm wrong, either.

[Shared404]

I hate that I have to feel agreement with this.

Edit {

Also, for those who are really concerned, I did test going to the landing page and disconnecting from the internet.

It still worked, which leads me to believe the creator was in fact telling the truth.

}

[victords]

Also, the EU is far from being this "digital warrior" some claim to be.

[offtop5]

Grad students working on Deep learning projects need to get data from somewhere.

[raxxorrax]

Putting all those images on a pillory on some marketplaces would certainly raise awareness for privacy.