3 ms·
> integrate an LLM into Slack They are already training their models https://slack.com/intl/en-gb/trust/data-management/privacy-principles https://slack.com/in
by surfingdino 2y ago
> integrate an LLM into Slack
They are already training their models https://slack.com/intl/en-gb/trust/data-management/privacy-principles https://slack.com/intl/en-gb/trust/data-management/privacy-p...
> Microsoft to integrate an LLM into outlook
Unlikely to happen. Orgs that use MS products do not want content of emails leaking and LLMs leak. There is a real danger that an LLM will include information in the summary that does not come from the original email thread, but from other emails the model was trained on. You could learn from the summary that you are going to get fired, even though that was not a part of the original conversation. HR doesn't like that.
> However, the more data you feed into an LLM, the smarter it should be in the response
Not necessarily. At some point you are going to run out of current data and you might be tempted to feed it past data, except that data may be of poor quality or simply wrong. Since LLMs cannot tell good data from bad, they happily accept both leading to useless outputs.
- geoduck14 2y ago>>Microsoft to integrate an LLM into outlook Didn't they already do this? A friend of mine showed me his outlook where he could search all emails, docs, and video calls and ask it questions. To be fair, he and I asked it questions about a video call and a doc - but not any emails, we only searched emails. This was last week amd it worked "mostly OK," but having a q/a conversation with a long email feels inevitable
- derefr 2y agoAsking questions about a document is one thing; asking questions that synthesize information across many documents — the human-intelligent equivalent of doing a big OLAP query with graph-search and fulltext-search parts on your email database — is quite another. Right now AFAICT the latter would require the full text of all the emails you've ever sent, to be stuffed into the context window together.
- bongodongobob 2y agoYes this is already a thing. Copilot for m365 fine tunes on your entire orgs data.
- surfingdino 2y agoSo that's a hard stop for many orgs. Not everyone is supposed to see all of the org's data even if they work for it.
- bongodongobob 2y agoThere are still controls on that, it doesn't give you access to everything. It took forever for legal to sign off on it.
- surfingdino 2y agoThe definitely added it to the web version n LinkedIn, you can see it when you want to write or reply to a message and it gives you an option to "Write with AI".
- derefr 2y ago> There is a real danger that an LLM will include information in the summary that does not come from the original email thread, but from other emails the model was trained on. You could learn from the summary that you are going to get fired, even though that was not a part of the original conversation. HR doesn't like that. There could be separate personal fine-tunes per user, trained (in the cloud) on the contents of that user's mail database, which therefore have knowledge of exactly the mail that particular user can access, and nothing else. AFAICT this is essentially what Apple is claiming they're doing to power their own "on-device contextual querying."
- surfingdino 2y ago> There could be separate personal fine-tunes per use Yes, but that contradicts the earlier claim that giving AI more info makes it better. If fact, those who send few emails or just joined may see worse results due to the lack of data. LLMs really make us hard to come up with ideas on how to solve problems that did not exist without LLMs.
- simianparrot 2y agoA problem looking for a problem to solve is how I like to think of it