Post

I would have thought that AI had made up the combi between 'airline' and 'service'... 🙄

Terrible. And just as much an issue for how Google Ads works as it is for AI.

The enshittification of the internet is slowly turning me into a Luddite. Not the technophobe meaning, the original definition.

This is why I have a browser extension that blocks the AI summary. I don't want to believe confident-sounding lies.

Same happened to a former coworker.

www.theguardian.com/world/2024/f...

Regardless, it's deeply frustrating that likely almost every single thing that someone caught and fixed in the old Google answers is going to have to be caught and fixed *again* because it's a new training set and model.

The scammers had my in-laws install remote admin software on their phones and then used that to keep texting my wife until the in-laws realized and shut their phones off.

While we are at it, a lot of blame goes to the companies actively HIDING their real phone numbers for years to try to make you use shitty chats so searching for them was the only way to get a person

It might also be worthwhile to report it to the FTC!

Same thing happened to me googling Apple Customer Service then they got my mom doing the same with Netflix Customer Service and it even had a hotlink to call them from the search. They ALMOST got me and got even closer to ALMOST getting my mom. How is this not regulated yet?

Takeout spots near me have signs warning there are no refunds on fraudulent orders. Fake ordering websites are ranked higher than legit ones and linked on the Google map listings; businesses can’t get them removed. Madness.

US Bank is the same. The have a number close enough to the real one where if you switch a digit you get an AI scam

Yikes 😬 So glad all the search engines are broken now! I was looking up a an ancient city to see how to spell the name and weirdly ai was the only one who would take me to the wiki link and you can’t tell me that’s not on purpose

That happened quicker then I thought tbh. It was only a matter of time before scammers would try to hijack AI (or as I call it Artificial Stupidity), but never in 7 moons would I've thought that it'd happen today with the still somewhat carefully controlled datasets...

United brings a lot of this on themselves by refusing to publish the info. Used to run a tool for internal United employees that had a support form. We had to turn it off because we got inundated with United customer looking for help with their flights. AI makes that problem worse.

This happened to me this past summer with Southwest, was a very sobering experience. Had I been like 10% less canny I easily could have gotten swindled.

How can they call it "intelligence" when it doesn't even check for basic scammer tactics?

Ai should not be called intelligent. That is beyond wishful thinking. It has no ability to quantify valued information verse low value. Just how much something appears. It’s close to useless if you can’t on accuracy it’s just further diluting and already waining clarity of quality info.

📌

yeah like legit scary, I've reached even higher levels of hating AI shit

Scammers were already figuring out ways to game the results, and insert their phone numbers into Google’s results. But AI is amplifying it.

This is such a huge problem that there are entire scam call centers now pretending to be US corpos. Like professional operations with hundreds of agents. Google fucking sucks.

How's that for a cursed timeline?

The only AI thing that genuinely scares me is the next decade of scams old people will have to navigate

Google needs to be legally liable for this shit

I know, I know. Sorry, didn't mean yours was wrong, just that when it happens at the airport it's easy to fall for it because it's a moment of rush and vulnerability.

There's all sorts of variations of this. I've seen them for locksmiths, every government service, NY's MTA's lost and found... Unfortunately at this point the only way to avoid this is to start at the official website and work your way through it: www.cbsnews.com/newyork/news...

Almost like entirely useless things only meant to make a few people rich are an open invitation for shady characters to profit off of it

…. OH COME ON

noooooooo omg (I am a programmer, I fully understand how bad that is. WILD.)

Google's AI told me yesterday that the resistor colors Orange, Brown, Yellow translated to a 1k value at 0.05% tolerance. It's 300k and, without the fourth stripe, it would be 20% tolerance. As it happens, the fourth stripe was gold, which is 5%. Where it got 0.05% is an unsolvable mystery.

AI isn’t capable of skepticism. Calling it “artificial intelligence” is a misnomer.

Especially considering that their monetization of the algorithm has moved the best results to page 2, averaging result 17 according to Cory Doctorow as cited in his enshitification presentations.

AI isn't built to be skeptical - it's at best a series of Markov Chains with some fancy statistics used to weight the links.

Gaming googles search results has been done for a long time, but it seems scanmers all too often end up at the top now. Especially if you Google stuff like tests/reviews in non-english languages, 8/10 first page results are AI translated/generated scams. With just affilite links if you are lucky.

An LLM can't really be trusted to accurately summarize the top search results credulously, much less skeptically.

There is a lot of red-teaming and black hat stuff happening right now. We’ve got to either get to attribution of data or have extremely hermetic build & training pipelines.

Google is an organized-crime firm.

Also all the generated sentences contains a source link 🤔

Happened to me with Delta last year. Scammers all the way down

The irresponsible way Techbro's toss their stuff out there is arming thieves with a whole new arsenal of harm. www.businessinsider.com/scammers-use...

I haven't had to adjust the sound on an ad in a while.

6 months is the time when it takes effect. Assuming it's not stopped by lawsuits.

The trick with that is they technically are. It's just that they crank the levels of the entire ad right to the limit. So while the show has one scene of a guy shouting, every ad is a guy shouting.

AI at this point in time is incapable of understanding (and thus critiquing) the information it assembles. Ultimately, information that is bad is just as valid to it as good information from reliable sources. A network of statistics doesn't make for comprehension.

udm14.com

This is distressingly common if they aren't buying ads to get to the top they are slipping it into support forums.

high value targets. I have a friend who is a middle aged *airline employee* trying to change a reservation and it took him more than 15 minutes to figure out that his call had gone astray...