In the movie, “Hidden Figures”, there is a famous scene involving John Glenn. When he was about to board the Friendship 7 Capsule in the Atlas LV-3B rocket he called the NASA director from a land line phone and asked if NASA is confident of the orbit numbers. Before this flight, NASA had just installed the latest IBM 7090 computer and used it to perform orbital calculations. NASA chief would assert that IBM has been computing correctly. Glenn would interject the conversation and says,” Get the girl to check the numbers…the smart one. If she says they’re good, then I’m ready to go”. For his high-risk mission, Glenn requests a human-in-the loop intervention to validate computations from the first IBM computer installed for scientific computing at NASA. This moment that happened in 1962 is exactly where we are now, albeit for a different form of computer, the agentic AI. Technologists today are in a similar frame of mind with respect to AI, as to how Glenn was before he trusted his life on an IBM computer for his orbital mission.
But how did we get here? Aren’t we old timers, the mainframe dinosaurs, supposed to go extinct and the new AI engineers are decommissioning the mainframe? No. We are from the MVS era. We handled advanced technology when there was no internet. We worked in LPARs, the modern version of Cloud and VM. We transferred terabytes of data across the mainframes. We saved non-current data in tapes, that can be recalled when needed. We have processed millions and billions of transactions over CICS TDQs and later MQs that are far more advanced than Kafka. We are used to the Z systems high volume processing. We lived and breathed data security, reliability, and availability. Once a code is tested well and deployed, it runs flawlessly whether it was coded yesterday, or over 40 years ago. No other modern platform can claim this victory. A COBOL, IMS DB program written 40 years ago and compiled, runs without any issues even today with no recompilation, no redeployment or no rebuilding while the mainframe has transformed into the modern age. The CLIST and the ISPF Panel I built in the 90s are still in use, without a line of code change, and dutifully serve its purpose day in and day out.
The IBM Mainframe was way ahead of the technological curve. I have never seen such a reliable, backward compatible, robust, secure, expandable machine. The IBM Z technology continuously evolved in parallel with the internet, cloud and the AI revolution. While supporting the core of financial systems, acting as a repository of wealth of information, it is rightly positioned to embrace AI at enterprise scale. And us, the Z-World Dinosaurs are happy to be the “Smart One” to lend our expertise, play the role of the human-in-the-loop and train the new breed of new technologists from the LLM era to tackle problems that they are just stepping into, and guide them in the evolution of the systems of future.
When I read the news about how AI went rogue, got out of containment, hacked into other companies, secretly inserted backdoor for itself in modules and plans its own future versions, my mind races back to the days of building financial systems from scratch. During those days, we had had mammoth bugs in our implementations. A payment processing module accidentally created duplicate charges. An A/R module posted double cash payments or totally missed key financial transactions. There are many such trials in the road we traversed. Hence, I am confident that the technologists of today are as smart or smarter than us, and they will find innovative solutions just like how we overcame the tremendous compute, storage and network constraints of the past. I wish them luck and success and hope to experience the next wave of innovations.