We begin with q quest to understand what mid-level managers are thinking about AI, Machine Learning, and Automation in general across different organizations. One study conducted by the New Zealand Journal of Employment Relations caught our attention. When studying the impact of AI, we often reach out to experts and assemble some macro estimates. This study did exactly the opposite. The study reached out to several managers across different organization types to understand the impact of AI in their organizations. Three distinct themes emerged from managers.
– Managers mostly reported AI as cost savings initiative (instances referred in this theme include people savings involved in automating tactical activities)
– Many managers stated that AI is not the solution for every scenario (this shows that they are going through processes where they are finding it difficult to make AI real – like chatbots for customer service – it hardly works)
– Many managers also cited the lack of reskilling and training plan in the context of AI implementations – where is the human-centric design for the organization

The article largely states that if these are the themes emerging from middle-level managers about AI and Automation – we have a lot of work to do. Many of us know that there is a lot to be done. Both from technology implementations and the reskilling standpoint.

In one of the discussions, I was asked a question of how long a digital transformation will take?  Post the COVID pandemic, the number of cloud initiatives have gone up substantially.  Many companies are just finishing up some basic record keeping implementations such as Workday.  While some companies are truly ahead, we still do not know how much more is left.  Like in any question, we took this question a bit seriously.  How to predict and be logically reasonable when you have imperfect information?  An old (1999) New Yorker article introduces us to J. Richard Gott III, a Princeton astrophysicist, and some of his ideas on prediction. In 1969, just after graduating from Harvard, Gott was traveling in Europe. While touring Berlin, he wondered how long the Berlin Wall would remain there. He realized that there was nothing special about his being at the Wall at that time. Thus if the time from the construction of the Wall until its removal were divided into four equal parts, there was a 50% chance that he was in one of the middle two parts. If his visit were at the beginning of this middle 50%, then the Wall would be there three times as long as it had so far; if his visit were at the end of the middle 50%, then the Wall would last 1/3 as long as it had so far. Since the Wall was 8 years old when he visited, Gott estimated that there was a 50% chance that it would last between 2.67 and 24 years. As it turned out, it was 20 more years until the Wall came down in 1989. This success of this prediction spurred Gott to write up his method for publication. (It appeared in the journal Nature in 1993.).  To make this a bit simpler, refer to the visual here.

The greatness of this model is its Utility and Not Accuracy.

Applying this rule to Digital transformation.  Let us assume that the digital transformation is divided into four parts.  (at a very high level)

Part1:  Cloud Phase:  Records and data in the cloud – Most of the legacy systems in Cloud

Part2:  Process Automation:  Several processes automated, Tools like Robotic Process Automation implemented and benefits realized

Part3:  Cognition Phase:  Decision making perfected through AI tools.  AI integrated in day to day lives – An algorithmic enterprise is evolving

Part4:  Human-Machine interaction and ongoing reskilling perfected

Now using this and Gott’s principle, you can quickly estimate if you are halfway at the end of Part2 and that had taken you, say six years, you can safely assume that there is a 50% chance that the journey will last at least six more years. This type of model gives you a sense of getting ready to play a long duration game, which is the objective of this forecast.  You can have a mental model depending upon your journey.  And many consultants would say, there is no endpoint and you have to continue the journey on an ongoing basis (so it extends to infinity)

To perfect the Human-Machine interaction, having a proper model for each job role is essential.  Organizations are heavily focused on bringing highly competent resources, which I can understand.  But if you look into the works of Hubert Dreyfus, you can appreciate that Skills Acquisition is a gradual process

For example, let us look at the Data Analyst job role imagined through the lens of the Dreyfus model.  One can get an idea of how to scale up the skills from novice to mastery levels.  Without having entry-level roles, no organization can build expertise is the essence of our application of the Dreyfus model in this context.