[nvch]

I'm waiting for the moment when all alive internet users will become AI-ified to test how to show them better ads.

It's possible that all the necessary data is already there. In cloud storage plus those intrusive DBs for sale.

Eventually someone will start selling accurate personas for $0.99.

[i_am_proteus]

I suspect that in the near future, Meta will deploy advertisements where AI-generated images of persons who resemble, but are not identical to, an individual's friends and acquaintances are shown enjoying sponsored products.

[accrual]

That's terrifying, thanks. I'm sure dollar signs are lighting up in someone's eyes somewhere.

[knowitnone2]

now you're just giving them ideas. thanks

[netsharc]

"Your grandfather loved using Harry's to groom his private parts!"

[RandomBacon]

He hooked up with Grandma, so it must work!

[amanaplanacanal]

It's easy to forget what hotties they both were when they were young.

[fortran77]

They sort of do that now, using pictures of a people in similar age/generation and interest groups (jocks, nerds, etc) as your friends in targeted ads.

[jerf]

The only thing stopping that right now is that that's pretty expensive at scale. The research about using AIs to "nudge" people, the research that AIs can far more accurately determine your tastes than currently-used profiling techniques, all that is already in place and it's obvious how to take the next steps.

Jury's still out on whether AI is actually going to be a net benefit to humanity, or if the AIs will be so firmly under the thumb of their owners that eventually the only rational thing to do will be to disregard everything that may have come from them because you can presumptively assume that everything coming from them is for the owner's benefit and not yours.

[watwut]

Considering the people in control of AI are effectively sociopaths with not much interest in anything other then self interest, chances that AI will be net gain for humanity in short term is super small. Its owners will use it the way they use their money and power in general.

[jezzamon]

We are?? Just they don't run LLMs, but they certainly use a lot of AI to target the right ads for us.

I don't think you need to run the LLM, or that you gain that much from doing it. AI is probably going to be on the ad side to support micro targeting, is my guess.

[jlarocco]

I think a more interesting question is:

At what point can we give an AI agent $100, set it free on the internet, come back in a week, and it'll have $1000?

[sitkack]

Because it killed 5 people for $200 each by shutting off their CPAP machines?

[bostik]

I know you're joking, but this doesn't feel too far off the mark in this world of late-stage capitalism run amok. Give it another 15 years and the bleeding edge[!] insurance companies are likely employing agents to go after clients who have become a net drain on their P&L.

The agents probably won't be doing that "themselves", but instead will be offering bounties (think: contracts) on suitably well hidden assassination markets. After all, as a machine AI cannot be held accountable for what is essentially a management decision.

I'm personally still waiting for the first country to go full Running Man to solve their prison overcrowding issues, and in addition to entertainment licensing deals also offer state-sanctioned gambling options to get a second bite.

[sitkack]

United Healthcare already did this with their AI powered death panels that would bulk deny claims, even if they later allowed them, the delta-t causes bonus subscriber terminations.

[jlarocco]

Air Canada already tried to claim their AI chatbot wasn't their responsibility.

https://www.bbc.com/travel/article/20240222-air-canada-chatb...

They lost, but it's amazing that they even tried.

[GoblinSlayer]

Why would it come back?

[cwmoore]

For your data

[progval]

Where would the money come from?

[skylurk]

Selling tokens for access to its reasoning ability? (just like the rest of us)

[jlarocco]

That's for the AI to figure out. It's intelligent, right?

[progval]

What can your unsupervised AI do to turn a 900$/week profit that someone else's unsupervised AI can't do for 10$/week?

[squigz]

I guess the efficacy of that would depend on whether you think a realistic profile of someone can actually be made from their online activities.

[sitkack]

This already happens, it is just the base model plus some in context data.

[A4ET8a8uTh0_v2]

I am personally looking into 'backing up' my individual personal model, but it is not as straight forward as just copy pasting context data ( not if you want sufficiently deterministic/convincing model ). It is a fair amount of things to consider ( and not directly available on mainstream APIs like chatgpt ). Yes, the bare minimum can be done easily, but easy stuff is not as reliable.

[knowitnone2]

What's an "intrusive DBs"...databases?

[inetknght]

West World, here we come!

[Philpax]

obligatory relevant qntm story https://qntm.org/perso

[GoblinSlayer]

I hate ads mostly because they are unbearable. If advertisers can figure out how to make bearable ads, I'm not against that.