Less than 8% of the workforce use Artificial Intelligence and/or Machine Learning tools (AI/ML). An extreme statement but adoption quotes from differing sources range from 4% to 40%. Considering that Artificial Intelligence in its many guises has been touted as a business savior for 20 years – any number less that 50% is incredible. The focus on the executive office with AI tools, while increasing their visibility (because of the target audience profile) has not been pushed down to middle management or to the field in most cases. The middle management group generally has adopted AI almost as a shadow – using Excel spreadsheets and the like to support their decision making. However shadow AI has significant challenges – info is not shared, not current nor vetted by business process owners (eg the CFO).
There are a number of reasons for the fact that automating business intelligence gathering and dissemination. Chief among them are:
- the lack of recognition of information and its use as a corporate asset
- lack of vision by corporate IT enablers to provide tools, access and capabilities to unlock the asset
- Short term vision of management where keeping information hidden has often been seen as a statement or demonstration of power.
The core problem in business is that the business continues to run with adequate tools and approaches – like it has for the last 50 years, but the world is changing rapidly around us. Information amounts, availability, acceleration of information types are all challenges to individuals trying to make intelligent decisions.
In many cases, AI/ML implementation adds little value to decision making. There are examples in the marketplace in which AI has significant impact on organization’s value.
Traditional approaches to AI/ML have created islands of information or silos. However knowledge of the business requires cross sharing of information. But that information must be made readily available – not in reports, static spreadsheets or inflexible data bases.
The recent banking crisis is a perfect example where isolated AI was the cause of a global financial meltdown. Models built to automate trading and bank Risk calculation were designed with no account for catastrophic failure of the system. The downside – in the extreme we saw, was not taken into account because most of the AI developers and model builders had never experienced a major economic downtown so catastrophic failure was not within their sphere of understanding. Text books and articles about the 1929 crash were viewed as antiquities because they were caused in a period of no automation. The belief was that the current systems, automated to prevent excess collapsed when the ‘perfect storm’ of failures pushed the models into unknown, untested waters. Had the AI/ML and model builders evaluated risk at its extremities rather than guessing what the downside boundaries could be (usually relatively small percentages from their ‘experience’ ) then the automation could have mitigated some of the collapse.
AI/ML – much vaunted terms – in the banking world – as a rule – did not look at ALL the information and therefore could only develop answers and approaches that were departmental or functional. Risk in Banking has to look at all risk factors in a holistic manner bringing all disparate information bases together to create knowledge.
How did the banking system get so over-leveraged with all the financial reporting, regulatory body and so on? Might be a way to convey it. Should we focus on a less lofty example?
Why are we still not there after 20 years? We have more advanced technology that is cheaper and ubiquitous. AI/ML vendors and the IT departments that buy / implement their products have not kept up with the proliferation of data sources, information uses, and complexity in business decisioning.
Can we find something that proves AI adds little value to decision making.
How did the banking system get so over-leveraged with all the financial reporting, regulatory body and so on?
Why are we still not there after 20 years?
©Trevelyan Group LLC 2019