Artificial narrow intelligence
The Digital Transformation is fuelled by digitalisation, i.e. data, and it results in huge data creation and availability. This is the fertile ground for Artificial Intelligence. In turns, artificial intelligence powers industry to leverage from data and monetise the benefit of the digital transformation.
Narrow Artificial Intelligence (“narrow” because is field and topic specific) has become to the digital transformation what the Moore’s law was for electronics: an engine of exponential growth.
Narrow Artificial intelligence can operate on 4 dimensions:
- inside a product to deliver functionality;
- inside the company to streamline processes, manage equipment (including proactive/predictive maintenance), quality control (analysing components and assembly),…;
- optimising the interactions among companies in the value chain (supply and distribution chain, predictive resource requirement, just in time, …);
- monitoring the operation/use of a product and feeding back the emergent intelligence in the whole value chain to improve the product and the future releases.
The existence of these 4 dimensions is what has lead to the name of AI hyperSphere (hyper because it goes beyond the 3 spatial dimensions). The crucial aspect is that these dimensions are interacting with one another, reinforcing one another. This is what is creating a self-sustaining accelerated evolution, exactly like the Moore’s law operating at reinforcing the production tools by producing better chips led to an exponential evolution.
These dimensions are acknowledge by industry also in terms of relevance, see graphic with:
- the relevance of enhancing current products -fourth dimension-,
- optimise internal Operations – second dimension- ,
- optimise external operations – third dimension-, and
- create new products – first dimension).
Narrow Artificial Intelligence is being leverage in particular for its value in data analytics, image recognition, pattern extraction and machine learning. These are possibly the four most important areas of application in the context of Digital Transformation since:
- data analytics acts on the huge amount and variety of data that would exceed the possibility of human management (at least in reasonable time),
- image recognition can leverage on digital cameras that provide digital images and artificial intelligence algorithm appropriately trained can already exceed human capability,
- pattern extraction is usually considered a high level intelligence characteristic in the human “mindspace” although in the humans pattern recognition is mediated by (or has a pre-requisite) an understanding of what is going on whilst artificial intelligence pattern recognition is more tied to the detection of certain repetitive occurrences so that actually in this area there is a complementarity between the humans and the algorithms,
- machine learning has been evolving rapidly rom the stage of training an algorithm through submission of huge sets of “cases” and correcting the outcome of the algorithm to the current technology edge where the algorithm can self-train leading to autonomous learning. This is, however, creating new issues since it may look like humans are losing control on the machine “thinking processes” (the reason why an algorithm returns a certain result) and the difficulty in controlling the outcome.