A large US bank approached a global IT firm seeking solution for a painful problem - 3 million USD annual losses due to frauds.
This is how it worked: a new customer takes a loan from the bank, and links the bank account details of another customer for payments without their knowing. The fraudulent loan seeker had access to the victim's bank details.
The global IT company built an integrated solution comprising of validation APIs that can validate the ownership of bank accounts, in addition to an artificially intelligence layer that, as a second layer, could detect potentially fraudulent applications and warn the back office.
While the validation APIs alone brought in a major reduction in fraudulent cases, the AI based warning system flagged over 53 % of the 1 million applications checked as potentially fraudulent.
While it is yet to be established how many of those 53 % applications were actually fraudulent, and how many of those 47 % flagged appropriate were indeed appropriate, the eye catching detail remains the fact that a half of the applications were flagged.
If one flips a coin and guesses, they would be right half the times. Is an AI solution predicting 53 % offrauds adding value? How Big Is The Problem
Deloitte reports that between companies who do AI really well versus those who don't, the return on investments vary greatly.
Average ROI on AI investments came to be around 4.6 % with many companies generating no or negative ROIs.
In last 4 years, banking and public sector have made the most incremental investments in AI. Travel, healthcare, chemicals and core industrial applications have been the next leaders in incremental investments.
Automotive and defence have been the leaders in absolute investment terms. With such critical industries making such investments in their AI initiatives, it becomes imperative that AI lives up to its promise of value and delivery. How Do We Fix This?
Even the most excellent brain surgeon cannot perform the surgery without an anesthesiologist.
AI initiatives are very similar. AI cannot perform by itself unless supported by the the right data strategies, processing and storage framework (cloud), and most importantly without the right AI strategy behind it.
While the senior management may recognise the value AI investments may bring, they must be willing to invest in getting the ground ready for all the above before their AI investments start to bear fruit.
It is not enough to have investments behind an AI strategy. The strategy must align to the overall business strategy for the management interest to persist.
Skilled AI people are not common. Even more rarer are training infrastructures that can deliver the quality manpower needed to deliver complex AI solutions.
The framework needs to be nurtured and augmented with the right tool set.
An AI model does as it sees.
Over 90 % of the world's digital data was created in the last 5 years, most of it either remains unusable to individual use cases, or leaders are still figuring our how to leverage what they have.
A good data source if not present, can be bought, or augmented. Which then needs to be enhanced with data processing pipelines, staging infrastructure in order to deliver value.
The value of data keeps reducing with time. Hence a mature Data Governance practise is vital to keep data sources relevant.
For most organizations, IT is still seen to be very low in the value chain, as a mere enabler, far removed from the core revenue generation functions.
While nothing can be far from the truth, this is a cultural problem and is not going to fix in weeks.
Until then if IT and AI investments have to continue they have to remain cost efficient. Cloud provides the greatest possibility for keeping IT lean.
A good cloud strategy needs to be in place for AI to function.
Join industry leaders already scaling with our custom software solutions. Let’s build the tools your business needs to grow faster and stay ahead.