Consumer AI vs Business AI, prioritising reliable and cost efficient solutions

Quick AI research, first post. I am conducting some summer research to understand the capabilities of AI and assess today's implementation possibilities to build my own tool.

Let's start by defining AI as a black box, i.e., ignoring the inner mechanics, and try to understand it and the different types of black boxes using an input-output methodology.

Consumer AI or General AI

A complex system, a behemoth built on top of several interlinked black boxes, a large black box containing many small ones. It provides answers, even if they are “hallucinations”. A hallucination is an answer that seems correct and well argumented, however is very far from the truth and based on a series of incorrect best guesses.

While sufficient for individuals and hobby users, it is very resource demanding and not considered accurate. The energy used by one query could power several hours of a light bulb on and relatively slow. You might want an answer within a short time to take prompt action as well as a very reliable results which is usually the case for businesses.

Business AI

Business needs reliable and cost efficient results even if the scope is narrowed. Therefore, they tend to build their own solutions, tailored and customised in house tools, partly for better data protection.

For different input-output applications, see the example in the image, which relates to the sectors where these are used. You should focus on the ones most relevant to your end goal or domain of expertise and sector.

Generic types of AI depending on use case:

🔸 Natural Language Processing

🔸 Geospatial

🔸 Time Series 

🔸 Graph

🔸 Computer vision

Resource: State of Data + AI by Databricks