Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
Google AI

Google Has a Big AI Advantage: It Already Knows Everything About You (theverge.com) 33

Posted by msmash from the flywheel-effect dept.
Google's expansion of Gemini's data access through "personal context" represents a fundamental shift in how AI assistants operate. Unlike competitors that start from scratch with each new user, Gemini can immediately tap into years of accumulated user data across Google's ecosystem. The Verge adds: Google first started letting users opt in to its "Gemini with personalization" feature earlier this year, which lets the AI model tap into your search history "to provide responses that are uniquely insightful and directly address your needs." But now, Google is taking things a step further by unlocking access to even more of your information -- all in the name of providing you with more personalized, AI-generated responses.

During Google I/O on Tuesday, Google introduced something called "personal context," which will allow Gemini models to pull relevant information from across Google's apps, as long as it has your permission. One way Google is doing this is through Gmail's personalized smart replies -- the AI-generated messages that you can use to quickly reply to emails.

To make these AI responses sound "authentically like you," Gemini will pore over your previous emails and even your Google Drive files to craft a reply tailored to your conversation. The response will even incorporate your tone, the greeting you use the most, and even "favorite word choices," according to Google.

Google Has a Big AI Advantage: It Already Knows Everything About You More | Reply

Google Has a Big AI Advantage: It Already Knows Everything About You

Comments Filter:
  • Seems to me AI should know a lot about real things. Not a lot about individuals. Guess we know what they are selling!

  • fb (Score:4, Interesting)

    by fluffernutter ( 1411889 ) on Thursday May 22, 2025 @11:42AM (#65395815)
    Doesn't Facebook know way, waaay more?

    • Re: (Score:3)

      by reanjr ( 588767 )

      Only if you use Facebook. Google's data collection is far more ubiquitous across the web.

      • It is more ubiquitous, but Facebook does the same kind of stuff. Their trackers are on many sites. They also create shadow profiles.

    • Re: (Score:2)

      by allo ( 1728082 )

      Probably. But while their model is nice, their hosted services are merely toys. Google stacks Gemini up against ChatGPT. Meta puts a chatbot into WhatsApp. You can only compare it, if Meta some day launches a serious assistant service.

  • Personal data + untrusted input + exfil vector (Score:4, Insightful)

    by abulafia ( 7826 ) on Thursday May 22, 2025 @11:42AM (#65395819)
    Yay, this combines all the ingredients for lots of costly, embarrassing mistakes or worse.

    People will get used to this and stop proof reading, just like we see people doing with code and legal briefs.

    And then the robot will refer to someone's partner by the wrong name, leak something important in negotiations, or a thousand other stupid, preventable mistakes.

  • Total nightmare (Score:5, Insightful)

    by Morpeth ( 577066 ) on Thursday May 22, 2025 @11:43AM (#65395823)

    Another reason to use other options like duckduckgo, or other browsers (I still like Firefox even with its missteps).

    The big problem with this too is that you reinforce echo chambers using your history, which isn't really a good thing, getting a range of perspectives even in search / AI results helps mitigate exaggerated beliefs or viewpoints (e.g. 'x is everywhere', 'y is always happening', 'everyone believes z')

    I also wonder how long it will be 'opt-in' until they make it the default and you have to 'opt-out'

    Total clusterfuck of an idea imo...

    • Re: (Score:2)

      by reanjr ( 588767 )

      "you reinforce echo chambers using your history"

      While this risk certainly exists, Google seems to be actively fighting this. In the past, I was better able to reword searches to either add or remove bias. Now when I intentionally look for bias, I get an "AI" blurb telling me my thoughts don't match the mainstream narrative. It's hard to tell for sure, but the results also seem to have this BBC-style "teach the controversy" framing.

      (as an aside: Why do I look for bias? Because sometimes I'm looking to see

  • But it still can't figure out I don't want constant changes to the gmail interface. Interesting. Another win for AI?

    • Re: (Score:1)

      by Anonymous Coward
      The Gmail development team has to justify their jobs somehow.

      • The scam of our time is the constant update cycle. These keep programmers in jobs and give them zero incentive to do anything right because there will be another update next week. It is also an utter productivity-killer and in a few years, someone will mention this scam and it will be revealed. Until then everyone suffers so these companies can justify having lots of programmers floating around doing mediocre work.

  • Now we can explicitly and complicitly train LLMs to impersonate us, as individual and unique humans, with near-perfect fidelity. What could possibly go wrong? /sarc

    • That's been something I've been trying to do for months! Also, plausible deniability: I did't type that, AI-me did!

  • It does not. But it does not know that. Also, even beginning to store data about me without my explicite consent happens to be highly illegal. (Yes, I am in Europe...)

    • It's trivial for a sub-entity that isn't legally Google and that doesn't operate in Europe to spy on your public activity and build the same sort of profile that Google would and to then sell that information to them. The data will be compiled about you and washed in whatever way makes it impossible to legally do anything about it. As long as collecting the data is worth more than the cost of collecting it and shielding themselves from legal liability, they will do so. Computers make it inexpensive to do so

      • Re: (Score:2)

        by gweihir ( 88907 )

        And you think that GDPR is not prepared for that? Well. It is. Google buying that data would just be as illegal as them collecting it themselves. Oh, and they would have to inform me within 30 days that they bought that data and where I gave consent to that.

    • This isn't storing data about you in the model, it's including it in the context. For example, you might say "here's a technical manual for context, now help me troubleshoot this thing."

  • Pretty much Rehoboam. https://www.harpersbazaar.com/... [harpersbazaar.com]

  • No it doesn't. False inference is not knowledge. (Score:2, Insightful)

    by Anonymous Coward

    The brutal truth about Google is that they actively add a lot of bad and false and untrue data to the profile they keep on you, including a lot of data that doesn't really exist, because they make stuff up by trying to infer things about users.

    The end result is that users are not being given a say in what is added to these profiles, and so they are always wrong and include data that leads to less sales conversions. Most of the profiles I've seen also include incomplete data that is probably best not public.

  • Log out of your Gmail/Google account and change IPs. Dump cookies. Keep them on edge.

    Your other option is to fool the algo by spamming it with random stuff. I research every medical condition I hear people talk about. I search for random products. I type in the names of sex toys and random extreme political parties. Sometimes, I search for incoherent gibberish like I was on meth, tequila, weed, and fent at the same time.

    But mostly, I stopped using Gmail. My friends thought I was insane and ridiculous. Well,

  • Even those waiting around for the antitrust trial to conclude are going to be disappointed.

    Google has already planned around the remedies:

    • Chrome divesture - irrelevant - browsers are irrelevant. Google is moving everyone to apps as fast as they can. Imagine reading this on the /. app right now - it's just a skinned browser. Get rid of Chrome all you want - apps are the new web.
    • Apple agreement. Apple is moving everything to SIRI. browsers will b irrelevant - especially default 'engines'
    • Data sharing. Goo

  • I mostly use google for it's search, email, maps stuff. Originally the tradeoff was I gave you info, and you gave me personalized results, and we paid for it with ads. Now the results are terrible generic stuff. I can just use a chatgpt type product and get similar answers for a nominal monthly fee and no ads. Why would I ever go back to google at this point? My search volume with google is probably a third of a quarter of what it used to be.

    I don't see much value in using google's AI product unless

  • You can build a massive horde of Personally Identifiable Information (PII) on all your users, perhaps the largest non-governmental database on individuals in the history. But if your AI product is shit, it won't help you.

    End users, after being exploited for their personal information for two decades by big tech, are getting wise. The exchange of personal information for free services is becoming less appealing to people. I hope the trend continues and we crush every corporation that wants to collect, proces

  • Unlike competitors that start from scratch with each new user, Gemini can immediately tap into years of accumulated user data across Google's ecosystem.

    And this site has been cheerleading Android adoption the entire time. Oops.

Slashdot Top Deals

"It doesn't much signify whom one marries for one is sure to find out next morning it was someone else." -- Rogers

Close