Source: https://front-end.social/@fox/110846484782705013

Text in the screenshot from Grammarly says:

We develop data sets to train our algorithms so that we can improve the services we provide to customers like you. We have devoted significant time and resources to developing methods to ensure that these data sets are anonymized and de-identified.

To develop these data sets, we sample snippets of text at random, disassociate them from a user’s account, and then use a variety of different methods to strip the text of identifying information (such as identifiers, contact details, addresses, etc.). Only then do we use the snippets to train our algorithms-and the original text is deleted. In other words, we don’t store any text in a manner that can be associated with your account or used to identify you or anyone else.

We currently offer a feature that permits customers to opt out of this use for Grammarly Business teams of 500 users or more. Please let me know if you might be interested in a license of this size, and I’II forward your request to the corresponding team.

  • 2 Nut November@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    22
    arrow-down
    1
    ·
    1 year ago

    Gang, I hate to tell you this but this is what we mean when we say “you are the product” especially with free offerings.

    But if you hate that I have a worse thing to introduce you to: the internet. If you respond to this comment, or any comment on any lemmy instance or other federated service or website or blog… your words can be consumed, copied and used to train whatever anyone wants. It is trivially easy to create web scrapers with just a bit of coding knowledge. These days it’s pretty easy to then use that data to train AI models. To a computer, it’s just data.

    Grammarly is a product where you give it bad grammar and it gives you good grammar. Grammarly, like many products, gets better over time when it can understand what went wrong so its teams can make it right. This can often include any text entered into the program. I don’t know the specifics but they should be outlined in the privacy policy. A company using data it already has to train AI makes sense, especially if it anonymizes that data. It may not be ethical given that users weren’t aware of AI at the time they accepted the privacy policy, but with american capitalism a company can change a privacy policy and you can opt out if you don’t like it.

    That’s why we all have lawyers on retainer to read and translate all privacy policies for all websites and applications we interact with in a daily basis. Right? That’s normal, right?

    I will say, could this support person have meant that an organization with 500+ employees get a custom AI model trained on only the organization’s 500+ accounts? Because that would be better, and likely more ethical too.

    If that’s not the case and any content you have put into grammarly is being used to train AI, then I guess it’s time to stop using grammarly then huh? But it’s also time to stop posting anything on the web, too. Oh, and don’t publish anything, ever.

    Or, you could go with the flow. This data is mixed with millions of other accounts… sort of like what happened when chatgpt trained on anything you’ve already put out there. The only real concern I could see is if you discussed a very specific thing or invented your own personal coded style of writing and used it so much that, among the millions of other users, dominated the corpus and skewed the training model. Say there are only 5 grammarly users and you are number 5… you keep talking about “procorpia” being “mass sledge”, generating hundreds of entries with thousands of tokens “words”. By contrast let’s say the other 4 grammarly users only used it a few times a month to send short emails. Now, after training, the 6th grammarly user mispells a word as “procorpia” and grammarly generares “procorpia is totes mass sledge brah”. Suddenly, your secret is out.

    If, on the other hand you speak the same broken english as the rest of us, you are probably fine.