BPM.com
  1. BPM
  2. BPM Discussions
  3. Wednesday, 31 July 2019
  4.  Subscribe via email
Ron Schmelzer writes in Understanding Explainable AI:

Most of us have little visibility and knowledge on how AI systems make the decisions they do, and as a result, how the results are being applied in the various fields that AI and machine learning is being applied. Many of the algorithms used for machine learning are unable to be examined after the fact to understand specifically how and why a decision has been made. This is especially true of the most popular algorithms currently in use — specifically, deep learning neural network approaches. As humans, we must be able to fully understand how decisions are being made so that we can trust the decisions of AI systems. The lack of explainability and trust hampers our ability to fully trust AI systems. We want computer systems to work as expected and produce transparent explanations and reasons for decisions they make. This is known as Explainable AI (XAI).


Do you agree? Will we never fully trust AI systems until we fully understand them? What about the time when we have AI systems designing AI systems that are completely removed from human architects?
References
  1. https://bpm.com/blogs/understanding-explainable-ai
  2. https://www.cognilytica.com/2017/12/20/ai-today-podcast-016-explainable-ai-xai/
Find me at BPM.com.
Accepted Answer Pending Moderation
Agree indeed a very topical subject for the accounting profession which has serious issues in addressing accountability with assurance with the many surprise corporate scandals / failures. The accountants should take responsibility to track the creation of data whether human or machine created. This takes us straight to the BPM discipline which in a simple structured way will ensure disclosure of activity. This must include AI and Intelligent Automation of which AI is just one aspect. Auditors / Accountants are well placed to understand just how any automation creates important data and this will require these skills to be used as build of any automation takes place. Automation to bring business benefits must have clearly articulated inputs to deliver the expected outcome and thus support tangible assurance on the data created. Accountants should be able to assist their business colleagues to readily achieve this and in language of business not in IT jargon which has scared business for decades! This should readily deliver Explainable AI and indeed any Intelligent Automation and this includes AI creating AI! BPM should be in play to monitor the end to end creation of such capabilities / algorithms?

Will the Accounting profession rise to the challenge and bring tangible assistance to their business colleagues in understanding all aspects of Intelligent Automation and AI remains to be seen?
Comment
Accepted Answer Pending Moderation
Artificial intelligence is becoming increasingly popular in many spheres of modern business. However, in its classical form of algorithmically driven IT solution, AI is extremely dangerous for business applications. The danger arises from a lack of understanding. Crucial technical and business decisions are increasingly made without proper human supervision and control. This situation creates deadly threat for business and for the public.

Vivid example is crash of Boeing due to faulty AI flight control systems. It evidently happens due to replacement of well understood and proven physical designs based on classical airflow mechanics with emulated AI controls. However elaborate and perfect these inventions could be, they still miss fundamental advantage of simplicity, transparency and reliability of nature. IT systems are imminently too artificial and fragile to compete with physical reality. Failure to understand this fact costs business losses and human lives. No doubt, this tendency is just beginning and we will see many more tragedies caused by AI in business.

BPM is a unique technology capable of protecting business and public from unsupervised AI growth. By using BPM analysis we can reveal hidden structures and patterns governing AI solutions and turn them into reliable and predictable systems available for understanding and mindful evolution towards fundamental business goals. Careful and thorough modeling must be crucial first step in development of any AI technology feasible for business applications.
Comment
Accepted Answer Pending Moderation
CEO, Co-founder, Profluo
Comment
  • Page :
  • 1


There are no replies made for this post yet.
However, you are not allowed to reply to this post.