Seventy years ago, Alan Turing told a gathering in London, “What we want is a machine that can learn from Experience.”  ( )

Operative words here are machine – learn – experience.

Fast forward to 2017, we have models, systems and (narrow) demonstrable machines which can learn from experience. It is almost given that simple, redundant and repetitive works will be replaced by such machines soon. But how about little complex ones?

I work as an analyst, a business analyst to be precise. A very large portion of being one entails knowing, logical reasoning, communicating, documenting and proposing solutions. Knowing the domain, market, regulations, organization, systems, processes, best practices, changes in technology, past and dependencies to state a few. Logically break everything to why-what-when blocks, connect the dots and arrive at how (solutions).

The list may sound exhaustive and quite complicated with lots of interconnected works that need to be performed.

On knowing:

To be clear, we are not expecting a machine to be in the realm of abstract, philosophy or the art. Though abstraction gives an edge, it is not a necessary condition to perform this job. A postulate – how many books, journals, reports, regulations, laws have we read in the past year or even the past decade. The answer will surprise us. What we consider knowing as ‘lot’ is in fact a tiny body of knowledge. With the current advancement in Machine Learning, Natural Language Processing, Artificial Intelligence, Analytics and Big Data, we can make a machine ‘know’ whatever I need to know to perform my role. This is the easiest part a machine can do– to know.

But what do we do with what we know?

We go to the customers with the (incomplete and less) body of knowledge that we have, we understand their businesses – the organization, processes, systems and then narrow down to what they want or propose what we can offer. Both ‘what they want’ and ‘what we can offer’ finally boil down to productivity – how to do more with less. It can be streamlining processes, proposing new processes and/or systems which will reduce time and effort provide insights of the organization using data to make them more profitable and the like.

With available technologies, we can make machines learn the processes of an organization. Today, they can understand what we speak or what we type in a way which we didn’t imagine a decade ago. Natural Language Processing makes it easier for people to interact with such machines. Add to the fact that machines can be made to learn multiple languages, it makes us look like a dwarf. IBM’s Watson can understand languages as diverse as Hebrew, Farsi and Chinese.

Application development of common systems (or for processes) contain three basic and distinct blocks. “What do I (organization) want,” is the requirement “What will I get” is the trade-off between cost-benefit and technological constraints and “How will I (IT) provide what you wanted” is the solution involving technology questions – the mundane code and test.

Many analysts mostly work on the interplay between one and two above. Logical reasoning and knowledge on impact and interactions of multiple systems will make a person effective and successful. No one can know multiple systems and that currently, is an impediment which is acknowledged. But a machine can change that. Imagine the tools of technologies that we currently have and breaking down these into sub-parts make it easy to think that every part of the work that is currently done by me can be replaced by a machine.

The other portion of of the work is documentation of all that is to be agreed and that has been agreed, in such a way that everyone understands it in only one way, i.e., clear, concise, descriptive where necessary and most importantly without ambiguity.

Having such machine take over my job has enormous advantages to any company. One of the most important thing that comes across is the change of vendor. When a vendor is changed, the knowledge (learnings from the past) is lost with people who leave. Documentation and knowledge transfer can only go to a certain extent in fulfilling the void. But imagine having a machine, which grows each day, learns by each interaction and will stay with the company irrespective of a vendor.

It may sound complicated but what is required is breaking the overall work – what I do everyday by every hour and every activity and then look at low hanging fruits which can be taken away from me.

Why I welcome it:

It is not about machines taking over my job. It is not a question of being happy or sad about it. It springs out of philosophical inquisitiveness in me. What we are need not be constrained by what we do. Our work need not entirely define who we are. Imagine the march of technology from fire, wheel to what we see on screen. Every time, we have invented something, it has resulted in loss of job in one area and the creation of job in another. But the current set of technologies do certainly pose a risk to a large number of jobs and therein lies our challenge to do even more complicated work which machines cannot do.

One of the human faculties is thinking. But how much time do we spend on thinking? Compare this to how much time we spend on doing. This skew toward ‘doing’ will inevitably tilt in favour of ‘thinking.’

Another advantage is of productivity. If $100 can do 100 tasks and if the same can be done with $10 with equal if not more quality, it is quite beautiful, yes beautiful. What we do with the remaining $90 is for us, the society to decide. Will it enrich corporations or can it be used to eradicate global poverty? How do we distribute such increased productivity? These are the questions that as a society we must answer together with corporations and government. But increased productivity is in everyone’s interest.

“Without continual growth and progress, such words as improvement, achievement, and success have no meaning.” – A quote (I think misattributed to) Ben Franklin.

(P.S.: The use of the word machine throughout is not a physical one but to differentiate from us, humans)


Leave a Reply

Your e-mail address will not be published. Required fields are marked *