A few sentences in, I was thinking that the article felt AI-generated, so I scrolled to the bottom of the page. There's no author listed, but there is this disclaimer:
"AI assists in refining our editorial process, ensuring that every article is engaging, clear and succinct."
One thing I hope we'll see in the future on these types of articles is the ability to view the original prompt. If your goal is to be succinct, you can't get much more succinct than that.
I think this assumes a very limited scope of how AI gets used for these. As if the article is a one and done output from a single prompt. I can imagine many iterative prompts combined with some copying and pasting to get an hour’s worth of copy in five minutes.
Feels like this title could benefit from clarification that 'Massive Attack' refers to the band and not the concept of a large scale attack; perhaps "Band 'Massive Attack' Turns Concert into Facial Recognition Surveillance Experiment"
> The band deployed live facial recognition technology that captured and analyzed attendees during their recent performance.
I think more drama has been created around this than is necessary. Based on the video, the real-time projected visitor's faces were not analyzed. They were simply shown with a random description flag attached, such as "energetic," "compassionate," "inspiring," "fitness influencer," or "cloud watcher." It seems to be an artistic provocation showing what a real people analysis could look like.
> It seems to be an artistic provocation showing what a real people analysis could look like.
I that cause they should have used descriptions like "gay", "muslim", "poor", "bipolar", "twice divorced", "low quality hire", "easy to scam", "both parents are dead", "takes Metformin", "spends > 60 on alcohol a month", "dishonest"
The fact that people were uncomfortable with simply having their pictures taken and shown without their knowledge gives lie to the idea that "You're in a public place—of course you have no right to privacy." It's great to be given the chance to face your principles.
I tried to create an art piece sorta like this once. Video cameras in two separate places in the world, hooked up to a monitor. Made to look like a mirror, only you realize you're looking into a completely different place. So if you and someone else walk up to it, it's like you in another dimension. I was told I couldn't bring it to a regional burning man event because "it violates consent" (because they didn't consent to being filmed). Despite their being no storage or recording whatsoever and it only being a live feed to another identical event. The organizers just couldn't come to grips with the discomfort they felt that there are cameras capturing your image. We definitely need more of these projects so people don't keep their heads in the sand.
Neat and all, but I'd be even happier if they flirted with the experiment of actually touring a new album, rather than serving as trip-hop's answer to Roger Waters, touring forever on the same 12 songs.
I'd say the same thing but I saw them on the Mezzanine nostalgia tour in Chicago, which was very expensive, and it was... not one of the best shows I've seen. I'd seen them a couple times prior and they were fine (I was both times surprised by the guest vocalists they'd managed to drag along on those tours). The Mezzanine tour though was like Spinal Tap's appearance on the Simpsons; "there will be no encore!".
I thoroughly enjoyed their Toronto show on that tour. To be fair it was the first time I’d seen them in concert so I didn’t have any points of comparison.
I also hadn’t really clued in to just how political they were until seeing their visuals, which I also thought added a lot. Surely not everyone’s cup of tea though.
It's been 15 years since their last original LP and over 20 years since the last album anyone really cared about (Google their setlists --- they play more covers than they do tracks from their last LP).
In the industry, that’s known as face (or facial) detection, which is a different problem than face recognition.
Face recognition means computing which individual from some other database of people a particular face belongs to.
There’s also face tracking — detecting a face in an image and then tracking the same face across subsequent images. Which is often implemented by using a face recognition approach, but without any predefined catalog of people — you just dynamically fill up your face database as faces appear in the image sequence / video source.
'Face detection' means it can detect faces. 'Face recognition' means it recognizes the faces. A specific example of the difference: license plate detection will detect the presence of a license plate; license plate recognition will tell you the number on that plate.
Are the faces even of audience members? Seems...gimmicky. The faces don't seem to react at all, and all are making almost AI movements. Many look artificial.
And it isn't identifying the people or anything. It's putting some meaningless adjective like "Resourceful" below them.
Have seen this headline a few times and thought it was actual novel, but instead it's just a surveillance gimmick. Put a bunch of generative AI face loops with bounding boxes and adjectives.
"AI assists in refining our editorial process, ensuring that every article is engaging, clear and succinct."
One thing I hope we'll see in the future on these types of articles is the ability to view the original prompt. If your goal is to be succinct, you can't get much more succinct than that.
I think this assumes a very limited scope of how AI gets used for these. As if the article is a one and done output from a single prompt. I can imagine many iterative prompts combined with some copying and pasting to get an hour’s worth of copy in five minutes.
What is sadly rather ironic is the author's first name, "Al" looks like AI when stylised in the article's font.
https://marketoonist.com/2023/03/ai-written-ai-read.html
How could you possibly tell? I've been playing around with AI detectors, putting in known all-human samples, known all-AI samples, and mixed samples.
The only thing it's gotten right is not marking a human sample as 100% AI (but it marked one of the AI samples as 100% human).
Having such a mark would be a witch-hunt for sure.
I think more drama has been created around this than is necessary. Based on the video, the real-time projected visitor's faces were not analyzed. They were simply shown with a random description flag attached, such as "energetic," "compassionate," "inspiring," "fitness influencer," or "cloud watcher." It seems to be an artistic provocation showing what a real people analysis could look like.
I that cause they should have used descriptions like "gay", "muslim", "poor", "bipolar", "twice divorced", "low quality hire", "easy to scam", "both parents are dead", "takes Metformin", "spends > 60 on alcohol a month", "dishonest"
God damn those are 12 great songs!
I also hadn’t really clued in to just how political they were until seeing their visuals, which I also thought added a lot. Surely not everyone’s cup of tea though.
But yes. They do need new material dammit.
I don't see evidence of facial recognition.
Face recognition means computing which individual from some other database of people a particular face belongs to.
There’s also face tracking — detecting a face in an image and then tracking the same face across subsequent images. Which is often implemented by using a face recognition approach, but without any predefined catalog of people — you just dynamically fill up your face database as faces appear in the image sequence / video source.
Recognition implies associating the faces with an ID.
Maybe I should keep them.
Have they done this again with an updated system?
It’s hard to explain the concept of surveillance and its effects to laypeople. And the corporations absolutely know that.
And it isn't identifying the people or anything. It's putting some meaningless adjective like "Resourceful" below them.
Have seen this headline a few times and thought it was actual novel, but instead it's just a surveillance gimmick. Put a bunch of generative AI face loops with bounding boxes and adjectives.