“I” vs “AI”: The Power of Judgement | Don’t Put All The Eggs In 1 Algorithm Basket!
“I” vs “AI” by Hosni Tahsin: The Power of Judgement | This partial prologue is contributed by Hosni Tahsin, Chartered Accountant and the head of financial consultancy specializing in internal audit, risk and financial advisory at Monstarlab Enterprise Solutions Ltd.
Disruptive innovations, AI and strategies – these three have always been the super hot topics in all things related to enterprise since Elon Musk made C-suite talents think ‘tweeting’ was the new corporate board meeting. You simply won’t find a conversation without getting a handle on these topics and when it comes to conversing with clients – you just can’t cut through the artificial intelligence buzz.
It was during a fireside chat last winter – an investor raised an interesting question on how AI performs in dealing with sudden market volatility. I took him back to the LJM Preservation & Growth history. It was a fund that prided itself on using sophisticated algorithms to protect and grow investments in even the roughest of markets. But in February 2018 their methods were put to the test and failed. Over 2 days this fund lost 80% of its value when their AI didn’t react quickly enough to a market that suddenly turned volatile.
As we stood there drinking our coffee, I went back to 87’s Black Monday, when the Dow Jones Industrial Average Stock price dropped nearly 1000 points in a matter of minutes due to unpredictable outcomes from the algorithms. The algorithms designed for speed and efficiency, reacted to a large sell order by placing additional sell orders. This created a feedback loop – a spiral of selling that drove the market down at an incredible pace unlike anything anyone had ever seen before. It was like watching a high-speed train derail in slow motion-fascinating, yet terrible.
I recounted, how this event brought about an urgent search for answers and reforms. The market did indeed ‘lose its mind’ or perhaps it was machines whose programming lacked any human caution of any kind. To this day, events such as Black Monday and Flash Crash are wake up calls for AI integrators and regulators. These histories help us reflect on just how crucial it is that AI be properly incorporated in to financial operations and trading. It raises questions about the need for safeguards and a deeper understanding of how autonomous systems affect global finance.
Couple months later, a curious volunteer at a workshop asked me about the biggest mistakes I’ve seen in financial tech. I told him about the Knight Capital disaster of 2012. A routine software update went horribly awry; within 45 minutes from when the market opened for business their systems were spitting out chaos. It occurred so quickly and ruined careers as well as the confidence shallow in automated systems. It’s an instructive reminder about how much faith we place on technology and machines.
Long story short: Don’t put all the eggs in one algorithm basket!
The upshot: Human oversight can’t be replaced even with next-generation experiences.
Here, at Monstarlab Enterprise Solutions, we implement a robust RPA and AI governance framework with stringent risk mitigation protocols and contingencies for each project. Over the years, we executed several RPA projects like automating transactions, streamlining workflows, optimizing supply chain operations, and providing turnkey ERP solutions.
Through these projects, we always found that our client-advisor relationship was built on personalized interactions and reliability. A few months back we helped a client safeguard their assets from potential financial losses arising from a glitch in an AI algorithm. The situation came forward during a routine health check-up and within two weeks, we not only resolved the issue by implementing fallback protocols and real-time error detection system but we also refined their risk management and AI governance framework with best-in-the-class policies.
So, how do you future-proof your operations against unforeseen AI glitches?
How do your team stay vigilant and respond to AI/ML challenges?