Hacker News Reader

Comment by paxys | Hacker News Reader

Standard internet outrage bait. There are no ads in the photos. The person uploaded their photo to Meta AI's "imagine me" feature which generates photos of you in exotic situations, and now the company is...putting them in exotic situations. That's literally what it is for.

View on HN

It's technically opt-in. You have to use their image generator, which requires agreeing to this "images appear in your feed randomly" feature implicitly. [0] You think you're just generating 1 photo, but you're actually signing up to a service that generates photos and adds them to your instagram feed. Even their help docs don't mention that nuance though. [1] It's not technically an ad, even though it mostly functions like one since the purpose of these images isn't anything more than bait to get the user to use meta ai more... I guess.

[0] https://www.businessinsider.com/meta-ai-face-images-instagra...

[1] https://www.meta.com/help/artificial-intelligence/imagine/?s...

Either I'm crazy or everyone else here is.

"I asked Meta AI for photos of myself and it started advertising Meta AI by showing me photos of myself."

How does this "function like" an ad? Why are we even using the word ad? This is not how ads work. By this definition what isn't an ad?

These are newly generated images, not promoted by you, shown in your feed. That’s the true “service” on offer here, images are generated for you and put on your timeline.

But the UI to sign up for this service looks like a “type in the text box, get an image” sort of service. It also doesn’t mention this instagram integration, other than requiring you to log in with a meta account.

You still get to make your own promoted images, so that’s marginally useful.

The main weirdness is it making its own images, on your behalf.

If you get value from random slop images with no particular theme, then cool. You can look at them as they show up.

But really, these are just throw away assets. It’s just meant to get you to click thru into the meta.ai upsell this links to.

It’s not quite an ad… because technically… it did something for you? There’s some implication that this random feed of useless images of you is… valuable. So it’s a service you accidentally walk into using, that just happens to redirect you to meta.ai when you click on it.

It's absolutely still an ad; ads sometimes include samples, like when they hand out small bites in the supermarket, those might be tasty ads, but they are still ads.

It's not an ad. It's someone trying to figure out what to do with the crap ton of GPUs they've been allocated and told to find some way to show metrics for usage going up, up, up, and the only way they can think to accomplish that is to get people to use them with zero interaction, since basically nobody cares to engage with the features otherwise.

Gotta say, this first batch of released AI features for the masses has been deeply underwhelming. Custom emojis and random pictures of me in random situations in some feed where the only reason I'm not scrolling past it is that it's a picture of me I don't remember being in hardly seems worth the valuations the stock market has been giving these things.

I suspect that if there was a way to clean up the metrics to see only what people actually wanted to engage with with these AI products, it would pop the bubble instantly. So much gaming the metrics by forcing it on people then claiming victory for the AI.

(I don't use LLMs for coding but at least that has a very clear value proposition. Search is nice now, but mostly because conventional search was ruined with malice aforethought... search AIs will follow soon enough. Enjoy that honeymoon while you can.)

It is still weird. Imagine you try out Photoshop online with your private photos. A few minutes later, you start seeing ads with those same photos edited in different ways. Even if those ads are visible only to you, it will still feel pretty creepy.

No, the person uploaded their photo to Meta AI to get a picture with some background, once. They then started seeing ads for MetaAI offering to generate more images of them, for a cost. Only the ads were still using their face from the photo they only uploaded once.

This is like using an online photo editing tool once, and then seeing ads to that photo editing tool show up on the internet, still using the image you up loaded that one time.

Except the ads don't show up anywhere "on the internet", they only show up on the photo editing site.

The photo editing site was "MetaAI", the display site was Instagram.

They may have the same parent, but are distinct entities.

There’s no cost to generate more images.

There's no cost to watch other shows on HBO, but the 2 minute promo they play for a different HBO show before playing the show you requested is 100% an ad.

Fine if you call it an ad, but it doesn’t change the fact that the feature is free of cost.

My time is not free. My attention is not something I've ceded control of to HBO to do whatever the hell they please just because I want to watch The Sopranos.

Also, their support pages say that this can be turned off:

> You can turn this feature off and delete your setup photos in Meta AI settings at any time.

https://m.facebook.com/help/messenger-app/1108543930466238

Standard hackernews top comment dismissal.

I’ve developed a habbit of automatically downvoting the top comment on most popular submissions because more often than not they are always some knee-jerk, often stupid contrarian reaction to the submission.

I don’t know when this trend started but I don’t remember it being this bad in the past.

People have been complaining about this for over a decade now. Here's some comments by pg himself, on the "middlebrow dismissal":

https://news.ycombinator.com/item?id=4693920

https://news.ycombinator.com/item?id=4397542

I guess a knee-jerk reaction demands a knee-jerk response?

I completely agree, and I'm baffled that so many miss the point, and/or subscribe to the original outrage. This is the same as someone uploading a photo, adding a filter, and later the application itself suggesting another filter, with a preview.

I do think that Meta, and others should be grilled to hell and back - but this particular situation is not noteworthy at all. Goes to show actually, how important PR is - the thing, and its public perception are completely separated, and so, needs its own management. Even if the thing in question is a tech thing on a tech forum.

Yes,

but putting that aside meta does give themself the right to use your face for ads and has used it and has won law suites about it before (through the cases I was aware of are quite old, i.e. before various right to be forgotten laws or GDPR).

So if they want to use your face to create an bot account to "increase engagement" they probably can do so without (legal) repercussions.

Yeah not seeing it either, both images say, only you can see this..

Feels more like it generated more than the user asked for and now it’s just showing those images in their feed.

What nobody mentions about "only you can see this" is that "you" refers to ordinary users of the service and there is no mention if employees are included in that grouping. It has happened before that systems secure from the public were misused by persons with privileges within the organization.

The obvious response is "well, yeah. So? They need it." but that's not how ordinary people, who don't deal with this daily, think. When they see "only you can see this" I think they take it literally.

The computer is always watching and sometimes so are the people running the computers.

Reply on news.ycombinator.com