Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Slack AI Training with Customer Data

slack.com

Post reply on HN
Print view View original

51–60 of 426 posts

Re: Slack AI Training with Customer Data

#51

Post by leoh »

nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

Yes, consider an existing LLM being given “shitpost-y” messages and asking it if there is anything interesting in there. It could probably summarize it well and that could then be used for training another LLM.

etc etc

Re: Slack AI Training with Customer Data

#52

Post by nyc_data_geek »

beeboobaa3 wrote: post #48
nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

More ignored than forgotten.

Wallpapered over?

Re: Slack AI Training with Customer Data

#53

Post by kepano »

tikkun wrote: post #3

Eugh. Has anyone compiled a list of companies that do this, so I can avoid them? If anyone knows of other companies training on customer data without an easy highly visible toggle opt out, please comment them below.

if your data is stored in a database that a company can freely read and access (i.e. not end-to-end encrypted), the company will eventually update their ToS so they can use your data for AI training — the incentives are too strong to resist

https://twitter.com/kepano/status/1688610782509211648

https://twitter.com/kepano/status/1682829662370557952

Re: Slack AI Training with Customer Data

#54

Post by chatmasta »

nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

Why shouldn’t AI be able to shitpost too? At the very least, and much more importantly, AI should be able to recognize shitposting.

Re: Slack AI Training with Customer Data

#55

Post by nyc_data_geek »

leoh wrote: post #51
nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

Yes, consider an existing LLM being given “shitpost-y” messages and asking it if there is anything interesting in there. It could probably summarize it well and that could then be used for training another LLM. etc etc

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.

Re: Slack AI Training with Customer Data

#56

Post by mayank »

zmmmmm wrote: post #43

> For any model that will be used broadly across all of our customers, we do not build or train these models in such a way that they could learn, memorise, or be able to reproduce some part of Customer Data This feels so full of subtle qualifiers and weasel words that it generates far more distrust than trust. It only refers to models used "broadly across all" customers - so if it's (a) not used "broadly" or (b) only…

> They need to reword this. Whoever wrote it is a liability

Sounds like it’s been written specifically to avoid liability.

Re: Slack AI Training with Customer Data

#57

Post by williamcotton »

nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

The sheer scale of data on the long tail. Sure, the head is already a trash pile and has been for decades now, but there is plenty of non-monetized information all over the internet that is barely linked to or otherwise discoverable.

Re: Slack AI Training with Customer Data

#58

Post by budududuroiu »

hu3 wrote: post #39
budududuroiu wrote: post #34

> We offer customers a choice around these practices If you’re so customer-first, and make it so easy to opt out, just make it opt-in instead. Oh wait, you’re just a lying pathetic corpo

Sadly, even Firefox is using opt-out these days. I feel we're going downhill with regards to privacy. https://news.ycombinator.com/item?id=40355982 > What Firefox’s search data collection means for you > We understand that any new data collection might spark some questions. Simply put, this new method only categorizes the websites that show up in your searches — not the specifics of what you’re personally looking up.…

Yeah these statements are always so benevolent, with the patronising undertone of “this is for your own good, for a better experience”.

SMH

Re: Slack AI Training with Customer Data

#59

Post by pyromaker »

>Our mission is to build a product that makes work life simpler, more pleasant and more productive.

I know it would be impossible but I wish we go back to the days when we didn't have Slack (or tools alike). Our Slack is a cesspool of people complaining, talking behind other people's backs, echo chamber of negativity etc.

That probably speaks more to the overall culture of the company, but Slack certainly doesn't help.You can also say "tool is not the problem, people are" - sure, we can always explain things away, but Slack certainly plays a role here.

Re: Slack AI Training with Customer Data

#60

Post by nyc_data_geek »

chatmasta wrote: post #54
nyc_data_geek wrote: post #46

Story time. I was at a VC conference last year and if I learned nothing else there, I learned how to spell "AI". Every single exhibitor just about had their signage proudly proclaiming their capabilities in this area, but one in particular struck me. They were touting the API integrations they could offer to train their "Enterprise AI"/LLM, and among those integrations were things like M365, Slack, etc. It struck me…

Why shouldn’t AI be able to shitpost too? At the very least, and much more importantly, AI should be able to recognize shitposting.

This is the crux of it, and where I'm wondering if I'm missing something. Can it, today? My understanding is it cannot discern reality from fiction, thus "hallucinations" (a misnomer because it implies awareness, which these probability models lack).
Post reply on HN