AI is the New Covid

By Paul Preiss

It was the 5th meeting in row, about 2 weeks ago, when I realized we were going to start each meeting, every blog, with AI, GenAI, and its derivatives. “So what are your thoughts Paul on the impact of AI to our (business, technology, and architecture)?” I blurted out, “This is like Covid… consumes our every waking moment, and is just as unfulfilling.”

Everywhere we look we see AI experts popping up. All of a sudden every architect in the world is scrambling along with their bosses. Every consultant is discussing ontologies, AI models, AI moats, data cleansing, developers being fired, regulation, the death of programming, the death of creative arts, and of course (per usual) the death of architecture. Hmmm well, I call bullshit.

No Mr. and Mrs. Consultants you did not learn to be an AI master in the last nine months. No Mrs. CEO your steel manufacturing company does not need an AI-assisted production line. No Mr. Programmer, you are not going to be out of a job unless you suck, but sadly probably not even then. Historically innovation of this sort creates jobs for us, not diminishes them.

We are in fact just doing what we love, following the marketers on their merry way to having us spend trillions of dollars on the latest and greatest TREND 😍. Hey, it’s your money. Spend it if you want to, but the fashion industry has a longer-lasting attention span than those in technology.

I said it about cloud… governments will get involved and commoditization will kick in and it will be cheaper AND better to have a bunch of our stuff on our own servers. Sovereign cloud is the thing. Turns out that once you pay for a resource it doesn’t cost as much to maintain it. Turns out Opex isn’t the panacea of every technology model… turns out we don’t want the latest gizmo to deploy every time. Turns out, not everyone can be trusted to use the cloud in the right way (cloud native is years off for most). But it is just a great way to make cloud vendors rich.

I said it about microservices… complexity will kill their value. Developers will abuse the machine. An architectural style does not apply to every architectural decision. But it is a great way to make the consultants and vendors rich.

And I will say it again. AI is deeply interesting *to specific use cases* and *within specific business contexts*. It can help us code better. It can help us learn faster. It can help us search faster and predict better. It can help with the boring and mundane (a little bit). And there are some utterly wonderful business cases out there! But not as many as you think.

If AI is generating all the text then why not have AI read it all? Oh yeah because then the AI becomes WORSE than us. It is like feeding it a diet without the right proteins, it will just get sicker and sicker. But hey it’s a great way to make the consultants and vendors rich.

I hear all the stories from the companies everyone is trying to copy. Both inside and out. And I can tell you that our technology stacks are NOT ready for AI. They are old, poorly integrated, inefficient, and governed by people with too little cross-stack education. It’s a bit like the streets in front of our houses. If we are really rich they look fine, but a couple of blocks down, you need a four-wheel drive to visit a friend. We are a house of cards built by the cheapest vendor. Anything that requires this level of rigor is going to go up like a bonfire.

Governments have already started to regulate, and it’s going to get better and better (I’m a fan). Soon you will need a license to develop an AI architecture, as well as expensive permits. Oh, the genie isn’t going back in the bottle but trust me this is going to get expensive fast. It is going to double or triple the complexity of development pipelines, creative pipelines, and of product delivery pipelines.

AI in the Life Cycle

The goal of any technology is to augment existence and humanity. The goal of our company is to participate in that ecosystem in a beneficial way. These are not always apparent in the day-to-day.

What architects should be doing is asking where will AI play in our corporate and technical lifecycles in ways that actually increase value (always in a measurable way). There are three big value areas in the lifecycle to consider:

  1. Innovation: how can AI help us find real innovation faster and prove it more thoroughly?
  2. Adoption: how can AI help us to adopt change better, faster, or cheaper?
  3. Execution: how can AI improve our operations in business technology?

These three areas are linked to the innovation lifecycle, the program lifecycle, and the operations lifecycle. If you don’t do those well already, AI is going to cause a lot more problems than it solves.

So What Should We Do

The answer to that is easy. Stay calm. Jump to the head of the hype curve and dump your massive AI spend into modernizing your stack so you can apply the winning principles when they emerge. Laugh heartily at the consultants telling you to get on board now. Ask them for a board-certified architect (not an accredited one though 😂) to talk to instead of a salesperson. Focus on excellence in critical business capabilities and their supporting technologies. Clean the hell out of your data. Make sure your technology workforce is disgustingly well-trained. Integrate modern approaches to technology investment and value management as a fundamental part of your business strategy. Create clear investment and decision records. Use lean business cases.

This is how we are doing it at Iasa. And how many of our large company architects are approaching it? Evaluate. Learn. Create good dependable use cases. Run experiments.

And dip your toe in carefully. Use rational and creative thought as to where ALL of your tech can make you money. And when you are ready, hit it out of the AI ballpark if you need it.

Also if you could not bring it up first at our next meeting, that would be groovy…

Also if you are looking to talk about real architecture, I will be teaching in Amsterdam on Sept 4th! for more.