Knowledge Management, Artificial Intelligence and Business Value

The intense media and capital expenditures of the leaders in the Artificial Intelligence space have generated an abundance of awareness and curiosity amongst the public. The business community rightly reflects and explores how much value can be unlocked with AI.

The expectation of business leaders is that there is considerable value in information captured in unstructured enterprise systems that currently cannot be harvested. The information then can be converted into knowledge, which than aid operations, finance, R&D, customer service, clinical research etc.

Regarding LLMs, these are statistical algorithms (transformer structures) that pack text into their sub-characteristics and determine the best statistical sequence of words but without a ‘real world’ context check, inevitably resulting in ‘hallucinations’ (Carlo Grazianı).

Several critics have pointed towards these inherent disadvantages or major risks (Ed Zitron, Nassim Taleb, Lynt Flyberg). Sam Altman stated in one of his podcasts that ‘ChatGTP results should never be used as a source of primary information.’

Whether ubiquitous use of LLMs is possible and under exploration, chances are there will be only specific business scenarios where additional value can be extracted.

Diverting from the focus on LLMs in AI or AGI, we return to the adage business question   ‘Do we make (enough) use of our internal and extended knowledge resources to create competitive advantage? The answer to this question is informed by the famous ‘Iceberg’ model representing the ratio of explicit versus tacit knowledge .

Explicit

Manuals

Reports

Databases

CRMs

Tacit

Local personal experience

Project team behaviors

Intuition

Shared stories

Depending on the Project Business Case , a KM solution may focus on or be a hybrıd how to manage explicit and tacit knowledge for optimal outcomes

A simplified business model to produce value consists of  a primary value work stream by departments, supported by Service Departments, both of which capture data in various formats (see below). Many potential nodes exist where department interactions (supported by data) can occur. Which nodes are relevant depends on a precise definition of the project Business Case such as real use cases A and B.

Case A

A global FMCG company concluded that their brand reputation is compromised by inconsistent quality of liquid detergents produced in their international plants. Causal factors are different specs for local raw materials, production machine differences and operator practices.   As a result, the KM project is focused on the ‘back-end’ of the value stream, to align Bill of Materials, supplier choice but also generating more knowledge on physics and chemistry of concentrated liquid detergents.

Case B

A major Oil and Gas Services provider realized that premier value can be added by providing technical solutions to the off- and on-shore technicians deploying on-sıte solutions for energy operators. Down time on operators’ facilities is very costly and a threat to the Service providers’ reputation. Due to the outstanding knowledge of the internal R&D and technical departments a KM solution was developed to deliver customized technical solutions over satellite, supported by on-line expert advice.  İn this case the challenge was to deliver deep internal expertise to remote locations in a way that allowed technicians to solve problems quickly.