Google is using AI to detect fake online reviews

by WarriorForum.com Administrator
9 replies
And they're blocking more than 170 million fake reviews using AI-powered algorithm. Google's blog post also reported removing and/or blocking:
  • 170+ million reviews that violated GMB policies
  • 14 million increase in detecting videos that violate GMB policies
  • More than 2 million attempts of fraudulent claims on business profiles thwarted

Google says its continued effort to combat fake content will benefit users, and of course protect local businesses.

Chime in.
#detect #fake #google #online #reviews
Avatar of Unregistered
  • Profile picture of the author AISEOUX
    That's hilarious.

    I've always taken the reviews seriously and get real ones. Now to be fair many reviews, especially early ones come from friends and family. But everything has become so review-based I'm not even mad at people that use fake reviews. I'm all for the small business turning a profit vs large corporations.

    I might take some heat for that statement but it is what it is.

    Good for Google but I think combating anything AI is never going to work. AI in inevitable and it gets smarter every day. There's certainly programs and ways to modify simple chats on ChatGPT to have a unique style of writing. Especially if it's tailored for a brand or anything like that. Good luck Google.
    Signature
    {{ DiscussionBoard.errors[11785403].message }}
  • Profile picture of the author CyberSEO
    So? Do not generate fake reviews. Use GPT to automatically create product ads based on their actual technical specifications. I thought this was obvious...
    Signature
    CyberSEO Pro - the ultimate AI autoblogging and RSS, XML, HTML, JSON and CSV import plugin for WordPress with support for OpenAI o1, Claude, Gemini, Llama 3, Midjourney, DALL-E, Stable Diffusion and more.
    {{ DiscussionBoard.errors[11786289].message }}
  • Profile picture of the author Moodesburn1977
    thats great news it just harms the product and the online industry
    Signature

    Lets build a online business by giving value and learning how to build a email list
    https://davidmitchell1977.systeme.io...alueonwarriorf

    {{ DiscussionBoard.errors[11786312].message }}
    • Profile picture of the author CyberSEO
      Originally Posted by Moodesburn1977 View Post

      thats great news it just harms the product and the online industry
      Actually, it's great news for those who predicted this move and used synonymizers (robotic non-GPT spinners) to modify their AI-generated reviews. This can be done in either manual or automatic mode, depending on the tools being used.

      This very simple practice makes it impossible to detect GPT-generated content, because it completely breaks a GPT flow. As a result the product reviews are recognized as human-written by GPT-detectors. Don't be lazy and spend a few hours manually creating a good synonym table for the GPT model you are using.

      A product review on the screenshot below was generated by GPT-3.5 and automatically post-processed by a synonymizer:

      Signature
      CyberSEO Pro - the ultimate AI autoblogging and RSS, XML, HTML, JSON and CSV import plugin for WordPress with support for OpenAI o1, Claude, Gemini, Llama 3, Midjourney, DALL-E, Stable Diffusion and more.
      {{ DiscussionBoard.errors[11786366].message }}
      • Profile picture of the author morisses
        Sounds like this won't save you if they all cooperate somehow (maybe using AI), which was also mentioned in some of the recent articles, and then track reviewers/buyer relationship as Paypal or any other financial site would, by "behavior" IPs etc. This may of course terribly backfire and hurt regular innocent users - just like Paypal, eBay etc have been doing for years anyway, with their empty bans for nothing, because you logged in after a month from another IP and bought something slightly different for a change... and they never restore access to your account because there is almost no support, bot decision is final. Except it'll get worse, if they will target reviewing. Over years it might actually cause the reverse, funnily enough - fraudsters will find the way, they always do, but many normal users will stop posting reviews to begin with, if they run into shitty bot-related difficulties once or twice
        {{ DiscussionBoard.errors[11793772].message }}
  • The point, which seems to have been lost in the conversations here, is that Google's using Deep Learning algorithms to flag fake reviews. They shouldn't be cracking down on AI-generated reviews, nor on AI-generated content, correct? After all, they explicitly stated that they don't care if content is AI or human-generated, as long as it isn't spammy -- I shared their announcement regarding this in this forum years ago. For example, if bloggers write in their native language their real experience in using a product and use locally running open source LLMs or online services like ChatGPT or OpenAI developer APIs to write in English (or any other language) about their experience in review format, then that wouldn't be fake, would it? So Google's algos shouldn't flag it if they're aiming for accurate performance from their algos to flag fake reviews, right?
    Signature
    Need Custom Programmatic SEO or GenAI Engineering Work Done? Drop Me an Email HERE ...
    • Chief Machine Learning Engineer @ ARIA Research (Sydney, AU)
    • Lead GenAI SEO Campaign Engineer @ Kiteworks, Inc. (SF, US)
    {{ DiscussionBoard.errors[11789192].message }}
  • Profile picture of the author robyn seely
    Using AI to spot fake reviews sounds like a smart move by Google. It's important for us as customers to trust the reviews we read online. AI can help make sure they're genuine.

    It's good that Google is using AI to find fake online reviews. This will help people trust the reviews and make better choices. Do you think this will make a big difference?
    {{ DiscussionBoard.errors[11793682].message }}
  • {{ DiscussionBoard.errors[11793886].message }}
  • Profile picture of the author IM Simple
    This sounds great. I'm so tired of seeing fake online reviews when I'm literally checking if the products/businesses I'm looking at are helpful.
    {{ DiscussionBoard.errors[11794216].message }}
Avatar of Unregistered

Trending Topics