Seems like everyone is talking about artificial intelligence these days. So how big of a role do you see AI playing in BPM in the future?
We devised such an approach years ago. I wrote about it here:
In order to do so, it is necessary to define your concepts and terminology clearly. In other words, you have to treat design like a science as opposed to an art form.
All the Best,
I would say AI (aka Cognitive Computing) will play a significant role in helping knowledge work. Especially when assisting the knowledge worker or augmenting resources with lower skill and knowledge levels. Also AI can help process / Case managers to identify important conditions, signals and patterns
AI will gain popularity over the next few years and certainly make its way into BPM, but maybe not in our current view of BPM.
AI will gain popularity because people like Mark Zuckerberg will increase awareness. He wants to create basic AI in a home assistant (like Jarvis in Ironman) for himself in 2016. I personally think it is great to see these type of projects as it creates awareness and also drives the development of the technology that supports it.
AI in BPM will evolve from what we currently do now by baking machine learning into processes (to predict when machine will fail or customers will churn) to cognitive computing where process learn as they go and adjust as they go. Jim’s comment has some great examples of how cognitive computing will change business models and business processes. Jim has been an early advocate of this and I highly recommend that you follow his blog and articles.
BPM itself will evolve as we move up the maturity curve for AI. Right now we use machine learning to make our “workflows” (I say that with caution to this audience J) smarter. We essentially embed basic AI in our processes. As we get more mature in the use of AI we will be “acting” on AI through processes, which means processes will be embedded in AI technology.
AI has been the talk for many years but with the recent Zekerburg announcement AI is gaining lots of attention. I have always seen the evolution leaning to AI and as we did with computers the transition will happen and we will adapt. The role of BPM practitioners will also change to help with AI design and algorithms to include BPM. I see that we will get to the point where AI will have the capability to assess and make changes to processes and implement the change.
As we grow and AI is seen in many processes my question would be the effect of AI on resistance? With AI slowly taking over processes the concept of resistance would surly start to be eliminated, of course this is dependent on the level where AI is at and the design capability and empowerment, but at the end there will always be a human Level that will make decisions on how intelligent he/she would want their AI.
The term "Artificial Intelligence" should be replaced by the term "Amplified Intelligence". This is a quote from "Cognitive Computing" by Peter Fingar, The quote itself is from Patrick Soon-Shiong of Nanthhealth;
"I don’t call it artificial intelligence. I call it amplified intelligence. We’re coining this term AI3, meaning amplified, actionable and adaptive intelligence. We’re going to enhance the cognitive capabilities of a human being who can’t recall information at the speed and depth that we need to make the right decision."
As far as case management is concerned it is the actionable intelligence that will speak volumes.
It seems that with AI capabilities for retrospective analysis of execution data, simulation of models (i.e. process itself, customer behaviour, etc.), complex treatement of various data sources, etc. + automated-system-design (thanks to Tim) + some other tricks we may reach the situation when process follows CX (i.e. each process instance adapts automagically to its particular customer) much faster than today.
Must say not wholly happy with term AI but nevertheless agree with Jim intelligence applied to processes will play increasing role in the interactions with users. For example working with a partner we created the intelligent questionnaire where as answers are given the process automatically presents the next appropriate questions and ultimately can present some conclusions based upon the responses. It is all about engaging the user giving a good experience even "fun"!
This is what we call a "living process" and we can see that the future could see expansion of this capability. I think one of the benefits could be making it even more difficult for the "bad guys" to manipulate business outcomes for self interest, corporate or personal.
Back in 1999, Douglas Adams was asked about the effect of Digital Disruption on the Music Industy, by execs hoping he would say "not too bad".
He explained it like this...
It'd be like a bunch of rivers, the Amazon and the Mississippi and the Congo asking how the Atlantic Ocean might affect them… and the answer is, of course, thatthey won't be rivers anymore, just currents in the ocean.
AI will have that effect on BPM. BPM will be just a current in the ocean which is AI, or more specifically, machine learning.
Machine Learning tests your processes in real time. Harvests all the data BPM provides and works out instantly what the optimum process is for the next user. It is like totally dynamic processes which self-optimise, adapt as needs change and continually AB test.
When you buy a car, the day you drive it out of the showroom is the best it ever gets. Not the same with, for example, a smartphone, which improves with every update. Processes are usually like the car - while you can change them, the change management processes are so onerous you really don't much, especially if the project team has long gone, clutching their fat cheque. Machine Learning is like the smartphone, only without the release cycle. It continually improves.
But it is better than that - it adapts too. No more surveys to see what people think - you can see what people actually do. If you have a possibility, it is the work of a moment to test it. And the system continually throws up new possibilities to test.
An AI system probably starts at around the same efficiency as a BPM process. But within days it is way ahead. Given scope, that curve is logarithmic.
The good news is that AI is still process, just dynamic. Process becomes more important, not less. And a human can often make the jump between tracks the computer cannot see. Just as the best chess players are not computers, nor people, but combinations of both, there is massive potential for BPM people who retrain as datascientists. But that window is already closing as more people migrate from other disciplines. Take the plunge and come on in, the water's lovely.
What's AI? Machine learning? Which is "analytics-writ-large"? Or "autonomous intelligence"? Which is the layperson's imagination as to what AI is. Or how about "automated autonomous decision-making"? Which is less ambitious and more realistic today.
All these things, whether in early stage or only in the imagination, are or may be excellent complements to business process, as noted by multiple correspondents here.
Except . . . two quibbles.
1. A FREE LUNCH? There isn't one. Domain knowledge and modeling (and likely ontologies) are an essential foundation for AI-enabled BPM. Even machine learning needs context. We found this out with expert systems in the 90's. What is the business case for acquiring and codifiying all that tacit knowledge?
2. AND WHY BOTHER? The ONLY reason for AI is to enhance decisioning (OK, "only" is a rhetorical exaggeration . . .) Decisioning maybe at the "edge" or decisioning maybe at a "macro scale". There's a big upside to better understanding decisioning in the context of BPM. AI functions to reduce the uncertainty for decisioning. A credit score or an dispatch choice or a reactor alarm are all about decisioning. What is the business case for better decisioning?