Leveraging Information and Intelligence

David Linthicum

BI and AI?

user-pic
Vote 0 Votes



Lori MacVittie and
I had had some twits back and
forth on business intelligence (BI) last week
, specifically its use within
an expert system.    While I'm not the
first to link BI and artificial intelligence (AI), at least conceptually,
perhaps the time is right to reconsider this.


I'm really talking about this in the context of active business intelligence, or the ability to monitor a business in real time, and also take automatic corrective action, typically without a human.   However, it's difficult to factor humans out of the equation when you consider that many decisions are "gut" calls, or are based on experience, and not some algorithm. 


So, what if you had a system that could take corrective action using years and years of experience from one expert, or perhaps the experience of many experts, all considered to solve a single business problem, such as when to stop using a supplier, or when to send a customer to collections.


In the past this has not been practical considering the cost of the AI systems, the ability for end users to set it up correctly, and just the whole science fiction nature of the thing.   However, today we have the links to most major processes within the enterprise, computing hardware is cheap, and expert system rules and logic could be delivered on-demand and setup automatically.   Thus, the practicality of mixing BI and AI is now there, I would argue.


Something for vendors to look into, and end users to consider.      


3 Comments

David,
You could argue that the 'embryo' of this exists through Complex Event Processing much of which is underpinned by the RETE algorithm, which itself underpinned a number of Expert System shells.

Part of what I think you envisage would require multi-paradigm learning. It is the elicitation of tacit knowledge that makes capturing the 'gut feel' logic somewhat difficult. Psychologically we may also skew the wisdom when a 'pure guess' pays off!

I certainly don't think this is a 'wild' idea, and with some of the CEP engines we have the bones of a 'rudimentary approximation'.

user-pic

To me, this is a de-coupling of BI and AI, where the AI is merely replacing the end-user of BI and taking an action based on the information provided to it.

My thoughts on a coupling of BI and AI is in how to deliver the correct information to the correct end-user (or other application) at the right time, from designing and building the full extent of the BI application to mining the data to discover patterns (etc etc), all through AI. There will have to be a collusion between these AI "agents" to quickly update best-practices based on new discoveries and technologies, almost like a network of AI BI agents (new term AIBI?).

I don't think we are close to this concept yet, and until further advances in AI are made, I doubt we will be able to do it. But it would be great if we could just get an AI app to go into a business, discover all the information that exists and is available, gather a list of requirements, design and build the application, and then spurt out the end product to the end user. It would then automatically monitor the source systems for changes and update the BI sysems accordingly.

If there was ever a project that was created to try achieve this, I would love to be on it....

We are a start-up specializing in bank performance tools. The vast amount of public data related to bank performance does make the combination of pattern discovery among a peer group and corrective action for the individual firm a near term possibility.

Industry expert Dave Linthicum tells you what you need to know about building efficiency into the information management infrastructure

David Linthicum

David Linthicum is the CTO of Blue Mountain Labs, and an internationally known distributed computing and application integration expert. View more

Subscribe

 Subscribe in a reader

Recently Commented On

Categories

Microsoft,

Monthly Archives

Blogs

ADVERTISEMENT