AI for clowns
Before we get to the main topic, let me ask you — Who is the clown here?
Somebody took it upon themself and reviewed Microsoft Copilot Terms of Use (effective as of October 24, 2025)
They noticed these two statements (in bold):
Copilot is for entertainment purposes only. It can make mistakes, and it may not work as intended. Don’t rely on Copilot for important advice. Use Copilot at your own risk.
WITHOUT LIMITING SECTION 12 OF THE MICROSOFT SERVICES AGREEMENT IN ANY WAY, BUT FOR THE SAKE OF CLARITY, WE DO NOT MAKE ANY WARRANTY OR REPRESENTATION OF ANY KIND ABOUT COPILOT.
which is slightly different messaging from the product page stating:
“What is a copilot? -A copilot is a conversational, AI-powered assistant that helps boost productivity and streamline workflows by offering contextual assistance, automating routine tasks, and analyzing data.”
And then you read the update from the earnings call that among other things, “Microsoft 365 Copilot has 15 million paid Copilot seats, up 160% year over year. Businesses can pay $30 per user per month for Microsoft 365 Copilot.”
Are 15 million customers paying $30/month for AI clown? Or did Microsoft acknowledge that Copilot is a multibillion dollar circus monkey?
Mind blasting.
While we are on the topic of mind blasting, I would like to address another theme appearing over and over in mainstream media, blog posts and social media.
Let’s talk about COBOL, the idiot’s piñata.
I know, nothing can be more boring but before you hit the ‘delete’ button and unsubscribe, let me entertain you, the ‘Copilot’ way.
In the past I occasionally mentioned this topic as in this post “When absurdity is mistaken for market prediction,” but a LinkedIn post which a colleague of mine sent me the other day prompted me to be more explicit.
COBOL is a programming language going to be 66 years old and still going strong. The name is an acronym for Common Business-Oriented Language. It is used by banks, governments, insurance companies — places where stability and reliability is a must. That’s why it is running mainly on mainframes.
As you can imagine, people who prefer software for its entertainment value can’t stand it. Accurately calculating your pension, investment or life insurance is so nineties.
Here is the link to a post written by a self proclaimed Senior Tech Leader.
The post starts strong: “Will Vibe Engineers save the dying COBOL application? No. But we will save its soul.”
You know that you are in for a ride when ‘vibe coding’ and ‘soul’ are mentioned together. I guess that’s how Senior Tech clowns get everyone’s attention.
After the few initial paragraphs we get into the step-by-step instructions “..how a Vibe Engineer approaches the ‘COBOL problem.’”
“1. We don’t translate; we extract.
2. We architect the “Vibe.”
3. We agentically generate the new syntax.”
Finally the genius approach is summarized in the final ‘Takeway.’
“The Takeaway:
Decades of hardware-software integration cannot be replicated just by moving code. Vibe Engineers aren’t here to keep COBOL on life support. We are here to extract the operational genius of the past and orchestrate it into the infrastructure of the future.”
You might notice in the amazing takeaway, this senior IT leader declared himself the undisputed leader of all the Vibe Engineers.
<sigh>
There is nothing wrong in coming up with new ideas, new programming languages or new technologies. What’s wrong is that just because there is an avalanche of new technologies it doesn’t mean that everything has to be rebuilt at the same speed. It is ok to change an application twice a day if the customer demands or tolerates it and the company has the resources to do it.
Despite the fact that COBOL is 66 years old, one can call it mature, it is regularly updated (the last update was in 2023) and running on IBM hardware with its current processor architecture supporting AI acceleration.
While the Vibe Coding clowns entertain us, they are probably not aware that this ‘ancient’ computer delivers 99.999999% uptime. That translates to 31.56 milliseconds of per-server annual downtime! These systems are engineered for continuous operation, with built-in hardware redundancy allowing for zero downtime during repairs.
But the idiocy perspires when we are told that: “we don’t translate; we extract” — “Instead of asking AI to rewrite the code, we deploy discovery agents to map the dependencies and reverse-engineer the business logic. We use the AI to read the ancient scrolls and tell us what the system does, independent of how the mainframe executed it.”
There is this myth that just by reading code and on occasion reading ancient scrolls (that’s why we still don’t know how to build the pyramids), you can reverse-engineer the business logic and later “We feed the extracted logic to our building agents (like Claude or Cursor) and have them write the new application from scratch, tailored perfectly to the modern infrastructure.”
Allow me a non-technical example why this approach is flawed.
Have you been to Chicago? Great city with so many things to see. Amazing architecture, museums, beautiful waterfront, parks, music. One place to visit is the Griffin Museum of Science and Industry and while there, let’s go to Washington Park and because it is a nice day, let’s walk there. A quick search using Google maps, you get your directions and you are on your way.
Google maps tell you how to get there. Do you know what Google maps doesn’t tell you? That Washington Park is one of the most dangerous neighborhoods in Chicago.
Reading the code provides you with some information, reading ‘ancient scrolls’ helps too. Truly understanding what the system does and especially why it does it takes time and effort. That’s what the clowns don’t comprehend.
The recurrent pattern? Be aware of vibe coding clowns who are trying to entertain you by swinging at their piñatas.