Paradigm shift… without a clutch

By | October 13, 2024

One sure sign that the hype of any technology is ebbing is when you see the media coverage and articles that are insightful, pragmatic, less hyperbolic and are not recycled content. In recent months, I have observed such green shoots related to AI. My gut instinct is that most of the enterprise have conducted one or more POCs (proof-of-concepts) or pilot projects and are now evaluating what can AI do for them.

Key challenges still remain. How do you calculate ROI and convince the bean counters about the perceived benefits of adopting AI? What are the use cases beyond a chatbot, knowledgebase and semantic search? How do we make sure that our data is secure and will not be used by the black-box platforms and AI companies for training their models? Who owns the IP? What regulations will be brought in that may derail our investments?

The ocean will be boiled, one pot at a time. But now that some sanity is settling in, let us shift our focus to adopting this wonderful new technology into the businesses and personal lives.

Paradigm has shifted…without the clutch

The most important disruption that AI has introduced is how we (should) think of software and automation. It is no longer about deterministic logic. It is about predictions. Rather than asking – how do we train a model to do this? We need to start asking – what can the model learn from my data and what can it (now) predict for me?

Data is the new code. Instead of having a multi-year tech-debt of bug fixes and enhancements, an AI model, trained or fine-tuned on your data, will generate the desired output (generative or predictive) as new data arrives. Risking a hyperbole, the impact of this paradigm shift is the tectonic shift that will create a massive disruption across business and technology landscape. Whoever gets a grip on this concept will create a competitive edge that will be very hard to surpass.

AI is not an island

When new technology arrives, the typical response is – how do we replace the old with the new? The assumption is that the new technology will do what the old one did, faster and cheaper. This is why most of the initial POCs and use cases appear to be stand-alone solutions. For example, replace the old routing based chatbots with generative AI based chatbots.

The way to think about AI is as for it to be an integral part of enterprise architecture and solution. This requires the enterprise and solution architects to think beyond databases, code and interfaces. For example, an inventory management solution should include where AI will be invoked to make predictions. Should it be at the time that inventory is received, binned, returns, shipped or, maybe, all of the above? What type of predictions would be desired? Should AI also read signals from other systems, both internal and external, e.g., marketing campaigns, manufacturing, competitor offerings?

You can call it the ‘cognitive layer’ of the architecture or, use the latest buzzword in town and, call the entire architecture as ‘cognitive architecture’.

Shift in power base

The shift in power from the techies (IT) to the business will be the slugfest to watch. The techies will not let go of the power that easily but they are bound to lose control faster than they think. With conversational, natural language driven user interface with the system that can learn and respond with updated information as new data arrives, will not require an army of developers to churn out enhancements as the business changes. The erosion of power will accelerate as the multimodal AI models mature further.

This will necessitate the merging or evolution of techno-functional roles. All technical roles, especially the role of enterprise and solution architects, are going to become much more strategic and business oriented. Universities should start offering some subject matter courses for STEM students. It will make them far more employable in the age of AI.

As the enterprise plans 2025 budgets, I would highly recommend going full steam ahead on AI adoption; or else, risk being disrupted by startups or sidelined by your competitors who took that decision.

Leave a Reply