Rethinking and Improving the Bucket Brigade
The first time I heard the term "bucket brigade" to describe accounting and reporting processes was from the book The Future of Accounting. That description accurately articulated how accounting information systems were connected into a flow which included lots of humans, lots of software applications, and lots of electronic spreadsheets to connect everything together.
But then I ran across the lean bucket brigade. Lean positions the bucket brigade as a "feature" rather than a "bug". (Image by Christoph Roser at AllAboutLean.com under the free CC-BY-SA 4.0 license.)
As I explained in this article, Partially Algorithmic Processes, There are three common mistakes made when trying to improve processes:
- Automating bad processes: Before you attempt to automate a process or workflow; you need to get that process/workflow dialed in.
- Don't understand process being automated: Attempting to automate a process that you don't understand is a recipe for disaster and will not work. Important subtleties and nuances will bring you to your knees.
- Excessive automation: Excessive automation is a mistake. Elon Musk learned this the hard way, admitting, "Yes, excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are underrated.” Others have made this same mistake.
There are a few additional things that you might want to keep in the back of your mind as you leverage artificial intelligence and attempt to "automate work".
- Digital first mindset: A paradigm shift is occurring. A "new normal" is being created. The territory is changing. A new map is necessary to understand the new territory. Trying to use your old map to understand this new territory will be unsatisfying. To understand why you need to have a digital first mindset, you need to increase your digital proficiency.
- Path to elegant simplicity is through complexity: There are no short cuts in life. Anyone can create a kludge. It takes hard work and you need to embrace complexity in order to arrive at elegant simplicity.
- Most people don't understand artificial intelligence correctly: The information is out there if you know where to look. Machines are not going to rule the world. You need to be able to separate the hype from the real capabilities. Artificial intelligence is a tool. You need to use the right tool for the job. Everything has pros and cons; strengths and weaknesses; capabilities and limitations. Getting this wrong can be very expensive. If you don't understand both sides of the equation; you don't truly understand.
- Traditional electronic spreadsheets are a stepping stone: The electronic spreadsheet is a tool. That tool as strengths and weaknesses. Because electronic spreadsheets don't scale; a new type of electronic spreadsheet can, and will be created, to supplement traditional electronic spreadsheets. Traditional electronic spreadsheets will always be with us; but a new "graph first" or "digital first" or "knowledge graph oriented" spreadsheet-type thing will be created. Think information Lego. It is time to decouple information from its costume. It is time to think beyond the document and give machines a chance to succeed.
- Difference between interpretation and understanding: Interpretation and understanding are not the same thing. Machines interpret; humans understand. Machines interpret information according to predefined rules. Those rules are defined by things like schemas, ontologies, theories, and rules that have been represented in some sort of machine-readable form. Machines interpret, they don't actually understand what they are doing. Humans understand information based on context, skills, experience, research, observation, and reasoning. Humans actually know the meaning of something. Humans may even comprehend something which is to understand something completely. Humans bring meaning to their ability to understand beyond information that is actually provided in the form of intuition and common knowledge that machines simply don't have. Humans can understand implications, "deeper meaning" from context, experience, skills, and knowledge that goes beyond the actual information itself. Computers are dumb beasts.
- Humans are underrated: Professionals are held responsible in ways that those not a member of a trade or profession are not. As a certified public accountant I get this distinction. Are computer scientists professionals? Will they be held accountable when their tools don't work the way they are held out to work? Personally, I think they should be. Humans will always be in the loop. Risk is managed via the mechanism of governance. Governance is done by people, not computers. Governance is the control framework that keeps a system coherent. Governance is how a group makes sure things are done properly. Humans provide governance; not machines. Trying to get artificial intelligence to reverse engineer business rules from legacy systems accurately is virtually impossible, a fools errand. To gain the full, original business intent is simply lost. Reconstruction of business logic in this manner has been tried time and time again, measured against time and cost, rarely achieves satisfactory results.
- Agreement matters: Meaning and understanding are not the same thing. Meaning is an agreement. Meaning enables understanding. Understanding validates meaning. There is a feedback loop between meaning and understanding. Without stable meaning, understanding collapses into ambiguity. Without understanding, meaning is inert, no uptake. Meaning is intersubjective, decided by a community of stakeholders for an area of knowledge a.k.a. domain of understanding. Understanding is individual.
You cannot have only one person in a bucket brigade. A bucket brigade is a collective trying to achieve some common objective. Also, a number that I hear over and over is that about 70% of information technology projects fail. I speculate AI automation projects will follow this trend or will fail at an even higher rate.
Additional Information:
Comments
Post a Comment