The Perfect Storm is here

?

Imagine that you are an IT services company that has a wonderful business going for the last twenty years, based on cost arbitrage between India and the west. Your main business model lies in working on projects that are billed on a Time and Material (T&M) basis. A large portion of this goes towards lower end developers — coders, database administrators, manual and automated testers, web developers and the like.

Imagine now, that there is a sudden invention that obviates the need for all these folks. The invention can write code, test cases, create database schemas and tables and even write documentation for the code generated — in short, can do all that these developers can do. If the bulk of the money you make through your T&M projects is through them, then how exactly would you feel about it? Unsure? Worried? Scared?

You ought to be. And the reason is because the hypothetical scenario outlined above is real, very real. The advances in deep learning and the explosion in Large Language Models (LLM) and Generative AI (GA) are the tip of the iceberg and the thin end of the wedge in a disruption that is going to hit, if not already hitting, all IT service companies. While the spate of layoffs that from big tech firms in the recent past in the US and elsewhere may have much to do with the global economy, I am willing to bet that the coming two to three years will lead to an increase in such layoffs.

Companies in the IT industry are akin to ships out there in the deep blue sea with the perfect storm approaching and nary a harbor in sight. LLMs are now able to not just amuse and entertain lay folks who use chatGPT to ask quaint questions to post on social media sites but they are also able to churn out code in multiple programming languages, write unit test cases and provide documentation among others. Lest someone think that these coding examples are possibly wrong at times and are only for low level codes like those for website development and simple tasks, let me illustrate with an example.

Even a year or so back, the task of creating code for identifying objects from videos would have been a fairly specialized task left to data scientists or AI/ML experts. No longer. A decently worded prompt on chatGPT can provide the code in multiple programming languages in a jiffy.

Scared now? Think of it this way. If you are an IT services company providing services to a Fortune 100 or Fortune 500 organization, with a billing of five hundred people a month, shouldn’t you be worried as to whether your customer would now be willing to relook at their entire IT strategy? Wouldn’t a smart IT head in those companies, decide, that all he requires is the middle management from the service provider — the data architects, project and program managers and business analysts who probably make up a small percentage of the overall contractor work force — and instead work with a few inhouse developers who can use these LLMs with some oversight?

Think of what it is going to do to your revenues which have hitherto shown impressive growth. I bet you are properly scared now.

Is it all doom and gloom then? No. It never is. There is always a lot of churn when a new disruptive technology comes onto the market. It happened with steam engines, the internal combustion engines, transistors, and the internet. And so, with this. What it would need though is for IT services companies to realize that the old days of cost arbitrage and sweet T&M deals are coming to an end.

Companies need to think about the value that they can add to their customers. There are some things that LLMs still can’t do. Like analyzing businesses, identifying and specifying problem definitions, and coming up with solution frameworks. You would still need humans to do that. At least in the short run. This implies that consulting firms like the Big 4 are probably going to be less affected than pure IT services companies.

Once the problem has been well defined, anyone can come up with the required tool/algorithm to be applied. It is in identifying which algorithm/approach to take that the value lies. It is like the joke about the lawyer who charged a thousand dollars for a two minute consultation. When the client protested and said "I could have looked it up in the book myself", the lawyer replied "Correct. But the money is for knowing which book to look into for the answer."

The perfect storm is here folks. Merely battening the hatches and bailing water out is not going to save the industry. Ships have to be redesigned for the new world. The future demands it.

#AI?#GenerativeAI?#GenerativeModels?#LLM