Claude is different from S3. AWS doesn’t need to rifle through your files to stay ahead of the competition or to mine them for business ideas because the core business is overvalued and rapidly commoditizing. AI labs, on the other hand, have an incentive to exploit every last drop of data they can lay their ethically challenged hands on. And they’ve already demonstrated that they will do this, even when it involves blatant fiduciary violations (see eg, Anthropic / Figma board member scandal).
I have never heard of a cloud service of ISP actively taking corporate data and shoving it into other users inbox. If my company has a proprietary solution to a math problem that nobody else knows about but the AI service that we have been working with has access to it and Billy Main Competitor ALSO has that service and asks "how would one solve this math problem" the second the AI spills the beans, the suits in the c-suit would lose it and sue the AI.
If however, our internal IT group didn't read the manual correctly and set up our cloud service access parameters incorrectly so that anyone who was trying to probe our defenses could access the formula, Jim from IT is looking for a new job.
> I have never heard of a cloud service of ISP actively taking corporate data and shoving it into other users inbox.
You'd never hear about if it the cloud service was taking your corporate data and using it for countless other things though. Insider trading, selling leads to your competitors, pushing ads to your clients/customers with offers of their own products, using trends in what your company and others are doing to guide development of their own products and services. Done carefully, you'd never be able to trace any of it back to them and it'd take a whistleblower for you to find out it happened in the first place.
I don't imagine any cloud service that has massive amounts of data which could easily make them a ton of money is just ignoring it.
Corporations are amoral psychopaths that don't care about anything except getting more money and power. We have countless examples of them lying, violating the law, and even murdering large numbers of people (including babies) to get it. What makes you think a corporation would ever choose their "reputation, trust and agreements" over money and power?
Which again, pretty huge fuckup on OpenAI's side to run that sort of security testing on 3rd party hardware, and not on a airgapped machine, kind of amateur hour to be honest. Again, same principle, don't give it access to stuff you don't want it to access.
How are you gonna handle the relations that span across individual chunks... if a later chunk refers something from 2 chunks before using `it`, rather than proper name, how will you handle that? Because at query time, that later chunk would not match.
Absolutely a novice in this topic, but I would imagine that by simply having sufficiently big chunks it's simply not a problem? You surely have enough information in like a couple of paragraphs to denote in vector space roughly what it is about. So that both chunks would get found by a vector search, and then whatever is the logic it may put the whole original text of those chunks into context, but in any case enough so that an LLM can "reason" about the references in-between the two.
Chunks can only be as large as the embedding model’s token limit, about 512-1024 tokens usually. Anything longer gets truncated.
Natural language processing could expanded references, but it starts to get tricky. Do you use Graph RAG, embed another version of the chunk that is distinct from the full text version, etc.. Another layer of processing and data to keep in sync if the source dan be updated.
512 tokens ≈ 350–400 words ≈ a long paragraph or two.
1024 tokens ≈ 700–800 words ≈ about a page and a half to two pages.
I would be very surprised if that amount of text is not enough to encode a general topic into the embedding (otherwise, what would be the whole point of them?).
So if there is a meaningful reference in C referring to A (assuming A-B-C consecutive 1-2 paragraphs), I would expect that the content of the two at least superficially resemble each other, and a vector query for one would return both. (And also, if I am including A in the context after retrieval, then I better give some context before-after as well -- and references tend to be local).
But feel free to prove me wrong, I'm mostly just guessing, not even an educated (in the given topic) guess here.
What member freakynit said nearby about chunks and relations between chunks, plus the storage and information efficiency problem: make some calculations about storing vectors - for paragraphs and for collections of paragraphs -, then compare the needed space with the original data...
Because you could have clever ideas about vectors related to more paragraphs related in the document structure - but that would multiply the vectors. The index can become much bigger than the corpus.
reply