Slack AI Training with Customer Data • Hacker News Board

Re: Slack AI Training with Customer Data

#421

Post by nyc_data_geek » Mon, May 20, 2024, 3:13 PM UTC

brookst wrote: post #184

nyc_data_geek wrote: post #55

Earlier quoted context omitted.

This assumes everything in the training data set is accurate. Sometimes people are wrong, obtuse, sarcastic, etc. LLM's don't have any way of detecting or accounting for this, do they? That output, then being used to train other LLM's, just creates an ouroboros of AI generated dogshit.

And yet human civilization has survived the fact that many humans are wrong, lying, delusional, etc. There is no assumption that everything in our personal training set is accurate. In fact, things work better when we explicitly reject that idea. LLMs do not rely on 100% factually accurate inputs. Sure, you’d rather have less BS than more, but this is all statistics. Just like most people realize that flat earthers a…

Output quality is low enough that the term "hallucinations" has been coined. Statistics doesn't appear to fix this.

Re: Slack AI Training with Customer Data

#422

Post by lars_francke » Tue, May 21, 2024, 8:51 AM UTC

hmottestad wrote: post #420

lars_francke wrote: post #344

I contacted support to opt out.. Here is the answer. "Hi there, Thank you for reaching out to Slack support. Your opt-out request has been completed. For clarity, Slack has platform-level machine learning models for things like channel and emoji recommendations and search results. We do not build or train these models in such a way that they could learn, memorize, or be able to reproduce some part of customer data. O…

Contact us to opt out. If you want to exclude your Customer Data from Slack global models, you can opt out. To opt out, please have your Org or Workspace Owners or Primary Owner contact our Customer Experience team at feedback@slack.com with your Workspace/Org URL and the subject line “Slack Global model opt-out request.” We will process your request and respond once the opt out has been completed. I guess that was w…

Exactly, yes.

Re: Slack AI Training with Customer Data

#423

Post by anonyados » Wed, May 22, 2024, 4:50 AM UTC

Ha, the perfect embodiment of how capitalism works. Use public or private data in this case to build models to profit. Even in a paid service where your data is supposed to be private, you are not even in control of how your data is used.

Re: Slack AI Training with Customer Data

#424

Post by rbalebako » Fri, May 24, 2024, 9:53 AM UTC

hackernewds wrote: post #199

j45 wrote: post #186

Earlier quoted context omitted.

I'm imagining a corporate slack, with information discussed in channels or private chats that exists nowhere else on the internet.. gets rolled into a model. Then, someone asks a very specific question.. conversationally.. about such a very specific scenario.. Seems plausible confidential data would get out, even if it wasn't attributed to the client. Not that it’s possible to ask an llm how a specific or random comp…

exactly. a fun game to see why it is so hard to prevent this https://gandalf.lakera.ai/

Gandalf is a great tool for bringing awareness to AI hacking, but Lakera also trains on the prompts you provide when you play the game. See the bottom of that page. "Disclaimer: we may use the fully anonymized input to Gandalf to improve Gandalf and for Lakera AI's products and services. "

Re: Slack AI Training with Customer Data

#425

Post by usernamevasile » Sun, May 26, 2024, 9:16 AM UTC

they are not the only ones doing it; see how Meta is doing it: https://www.linkedin.com/pulse/open-letter-meta-vasile-tiple...

Re: Slack AI Training with Customer Data

#426

Post by modriano » Wed, May 29, 2024, 8:44 PM UTC

IMTDb wrote: post #89

koolba wrote: post #61

> We offer Customers a choice around these practices. If you want to exclude your Customer Data from helping train Slack global models, you can opt out. If you opt out, Customer Data on your workspace will only be used to improve the experience on your own workspace and you will still enjoy all of the benefits of our globally trained AI/ML models without contributing to the underlying models. Why would anyone not opt…

> Why would anyone not opt-out? Because you might actually want to have the best possible global models ? Think of "not opting out" as "helping them build a better product". You are already paying for that product, if there is anything you can do, for free and without any additional time investment on your side that makes their next release better, why not do it ? You gain a better product for the same price, they ge…

How could this make slack a better product? The platform was very convenient for sharing files and proprietary information with coworkers, but now I can't trust that slack won't slip in some "opt out if you don't want us to look at your data" "setting" in the future.

I don't see any cogent generative AI tie-in for slack, and I can't imagine any company that would value a speculative, undefined hypothetical benefit more than they value their internal communications remaining internal.