Playing devil's advocate here for discussion purposes:
Isn't the 1st step of any AI to collect as much pertinent data as possible? The potential I see in Palantir is that they have the base infrastructure covered (talent, tech stack, business relationships) to connect to all sorts of disconnected database sources.
The 2nd step will be to analyze that data for insights using opensource or in-house engineered algorithms - not Palantir's role right now.
Ex: My past firm played with Datarobot's AI solutions which tried to outsource data scientists as "consultants" but did not work very well due to lack of data engineering/domain expertise.
Could this be Palantir's next move?
Note: This may sound mildly speculative, but trying to approach it from an engineering and business perspective.
I’m not aware of any Palantir contracts that let them use, model with, or analyze data outside of designated purposes. It’s pretty hard to audit cloud services that claim to have isolated dbs, but it seems like such a massive breach of contract if they were actually doing this.
It’s the same risk if Microsoft cracked open all the sql dbs to analyze and train AIs. Or if AWS tried to use data stored with them. I’m sure it would be useful but would break every contract they have and since they are trying to make money off cloud services that’s a big risk for minimal return.
So unless there’s more evidence just saying something is possible is a rabbit hole for every cloud provider, right?
Isn't the 1st step of any AI to collect as much pertinent data as possible? The potential I see in Palantir is that they have the base infrastructure covered (talent, tech stack, business relationships) to connect to all sorts of disconnected database sources.
The 2nd step will be to analyze that data for insights using opensource or in-house engineered algorithms - not Palantir's role right now.
Ex: My past firm played with Datarobot's AI solutions which tried to outsource data scientists as "consultants" but did not work very well due to lack of data engineering/domain expertise.
Could this be Palantir's next move?
Note: This may sound mildly speculative, but trying to approach it from an engineering and business perspective.