Newsletters

Optimising financial processes

Posted on:

Digital Transformation – Automation or Augmentation?


Like you, I am engaged in an increasing number of conversations (unsurprisingly!) about the role of Digitisation, Automation, AI and Generative AI in Transformation.

It is common to conflate hot topics and themes, but it is helpful to unpack them.

We know that the demand for business transformation is generally driven by a need or desire to drive revenues, expand market share, develop long term profitable customer relationships, create and deliver winning products & services, maximise the return on employed talent and optimise costs and cash.

But, we tend to jump very quickly into how and what technology we can exploit to deliver the transformation.

Of course, we know that technology alone cannot do any of these things.

Technology is a “force multiplier”.

If the force is suboptimal or negative, this multiplication can have unpleasantly dramatic effects, as we see only too often.

“The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency.

 The second is that automation applied to an inefficient operation will magnify the inefficiency.”

So, our first responsibility is to define the efficient operation, the “end to end process” and the desirable, discrete “journeys” (there are always categories and segments of our customers, production and spend that need different routes).

The process and journeys describe and define the tasks required to deliver the desired outcome of an end-to-end process or value stream.

We need to decide how to optimally integrate, or string these tasks together, across business functions and locations.

Then decide how to apply appropriate technology.  

But it is a nuanced decision about the optimal role of human and machine.

Rory Sutherland, long time WPP executive, Vice Chairman at Ogilvy UK and author of “Alchemy: The Surprising Power Of Ideas That Don’t Make Sense”, wrote a brief, but fascinating, piece called The Doorman Fallacy.

This should be a must read in every business . . . 

It is the “iceberg principle” writ large . . .

 “The Doorman Fallacy is what happens when your strategy becomes synonymous with cost-saving and efficiency; first you define a hotel doorman’s role as ‘opening the door’, then you replace his role with an automatic door-opening mechanism.”

“The problem arises because opening the door is only the notional role of a doorman; his other, less definable sources of value lie in a multiplicity of other functions, in addition to door-opening: taxi-hailing, security, vagrant discouragement, customer recognition, as well as in signalling the status of the hotel. The doorman may actually increase what you can charge for a night’s stay in your hotel.”

Every process and constituent journey (and task) is an iceberg.

Consider what is below the waterline!

We have, in our arsenal, technologies that automate and technologies that augment.

These are both powerful concepts, but we should be clear about the difference.

Technologies that automate repetitive tasks are important. This makes sense because automation of repetitive tasks is a Boolean logic challenge.

AI, and specifically Machine Learning (of which GenerativeAI is a subset), apply algorithms and statistical analysis to large bodies of data. We need large bodies of trusted, quality data for this, but that is a discussion for another time.

A fascinating and illustrative example of this is in the field of pathology for cancer diagnosis.

The conventional process is that a blood sample from a patient is provided, and a drop put on a slide to be examined under a high powered microscope by an expert pathologist.

Given the scale of this activity, it was seen as a great AI opportunity to learn from large bodies of data.

An AI algorithm was developed to identify cancer cells in blood samples, trained on a large volume of data.

This resulted in the AI identifying 92.5% of cancer cases.

Impressive!

However, it turned out that diagnosis by a human expert pathologist identified 96.6% cases.

The most interesting element was that, when AI was used in conjunction with a human pathologist, a massive 99.5% of cases was identified.

WHY?

Whilst the AI algorithm never lost concentration and was consistent, the human pathologist was also aware of “edge cases” outside the training data and found potential anomalies in addition to the obvious.

This is a great call to action for the expert “Human in the Loop”.

Augmentation beats Replacement for complex tasks . . .

As the pathology example indicates, we should consider predictive AI for specific inputs to augment human judgement for next best action and expert decision making.

We should consider Generative AI and Large Language Models (LLMs) where a conversational interaction is desirable with an employee or consumer, where the textual narrative is beneficial (think hard about that!), whilst recognising the critical importance of the (hopefully) trusted data behind a very plausible textual narrative.  We have all experienced the GenAI “hallucinations” . . . 

AI, whether Generative or Predictive, is about maximising the effectiveness of the “Human in the Loop”. AI is optimised for augmentation.

Should Our Business Transformation be driven by Automation or Augmentation?

BOTH!

Effective transformation requires both Automation and Augmentation based on a sound understanding of the most effective end to end processes, journeys and alignment of tasks.

You can listen to Rory’s perspectives, including his observations on the “Doorman Fallacy” here . . . . 

Thanks for reading . . . .