Larry Ellison's latest craze: Vectorizing all the customers If you're an Oracle customer – throw a pebble into a crowd of 100 CIOs and you're bound to hit one – then Big Red has vectorized you. Or, more accurately, it has vectorized your data, according to Larry Ellison, co-founder and CTO, who lobbed about the terminology in this week's conference keynote as if it conferred some sort of …
Agreed - I think he is talking about data about Oracle customers and is talking up who AI can be used to find trends in purchasing. I would have thought Larry would be aware he could also do this with "SELECT * FROM CUSTOMERS WHERE ...." statements, but is clearly better at selling people things they don't really need than I am so I will defer to his superior sales approach.
Probably depends upon who the “customer” is, but this ability to vectorise data was possibly a consideration in hosting TikTok on Oracle cloud, given it would seem the customer was Trump and his backers….
I wonder if the Israeli’s are now using this for the Gaza phone data they were forced to move off Azure…
Somebody sued Oracle and Salesforce for EUR 10 billion in ca. 2016 for GDPR violation for a very similar data sharing case, albeit without AI. A Dutch court dismissed the lawsuit in 2022.
https://gtm360.com/blog/2024/11/27/b2b-data-sharing-by-saas-ai-whose-data-is-it-anyway/.
Let's see what happens to the next GDPR violation lawsuit, if and when one gets filed.
It will burst soon, I guess.
> Big Red wanted to ask its "reasoning models" what products customers are likely to buy in the next six months.
Based on the Oracle comms I see, this works about as well as "other customer looking at this product also bought *this shiny new thingy*, one is used to see on Amazon and the like, or the ton of refrigerator ads you see for 3 months _after_ you bought a new refrigerator (and forgot to clear the cookies out of your browser).
So, they basically achieve the same, often pointless, outcome by different means, e.g. by "vectorizing customer data" with much higher investment and much higher energy cost for AI to process their vectors. And with much higher environmental impact.
I must admit that's a somehow cool approach and all that, but pretty pointless from an economic and outright dangerous from an environmental perspective.
Not that I want to suggest there's some sort of coordination - other than fiscal gravity - but son-of-Larry is CEO of Paramount Skydance which owns CBS and is attempting to acquire Warner Brothers. So it's not just your corporate data that's potentially in scope for the Ellison dynasty.
We seem to be entering a new era of Rockefellers and Hearsts.
"The first thing that Oracle did [was take] private data and [make] it accessible to AI models. We took all of our customer data and we vectorized it." Hmmmmmmmm if they took their own data about their customers and waved some dead chickens at it to invoke the appropriate daemons fine. Stupid expensive way to achieve the outcome, but fine.
But if they used customers own data, without consent, permission or notice then that's an intriguing admission from Larry. Which opens up to possibilities, buried in the Oracle T&Cs is permission to do just that, probably phrased as training, teaching or product improvement language and we pinky promise not to share your data outside of Oracle. Or Oracle is taking a Meta approach to data slurping and saying eff it, we own the world, what are you going to do about it, never look in the rearview mirror, all your data are belong to us.
I'm betting it's door number 2 Monty
@AC
Vectors are a way of representing data (e.g. text) numerically (typically lossy)
Used a lot in LLMs a way to help pattern matching.
Massively simplified but..
Say a question is asked of a chat bot and that you have, vector data stored for entries in your company support / helpdesk data.
If you analyse the question and produce a vector, then search your support / helpdesk vector database for closest matching vector, then with a bit of luck you may get a relevant document returned.
Sort of thing where a chatbot would say - we found this in our database is it of any help and then present the document / snippet from best match (or often will give multiple "hits" based on how close a match you define as acceptable)
