The AI Revolution Won’t Happen Overnight — And That’s Exactly Why You Should Pay Attention
- Asma Asad
- Sep 2, 2025
- 2 min read
Everyone’s talking about AI like it’s already redefined the world. Headlines scream about trillion-dollar impacts, record valuations, and productivity boosts. But, according to Paul Hlivko, EVP & CIO at Wellmark and author of a Harvard Business Review article, we’re getting six big things wrong about AI’s value, adoption, and speed.
AI will change the world, just not the way we’re being sold.
AI’s real value won’t come instantly
Like electricity or the internet, AI is a general-purpose technology (GPT). These take decades to deliver returns. Despite big investments, AI’s productivity impact so far? Minimal. Most users spend under 5% of their time on GenAI tools. Why? Transformation is hard, integration harder, and hype outruns infrastructure.
We’re too optimistic about enterprise adoption
Consumer use ≠ enterprise reality. ChatGPT might feel magical, but embedding AI into legacy systems, meeting compliance, retraining teams, and redesigning workflows? That’s grind, not glamor. Even IBM Watson — once promised to “outthink cancer” — collapsed under real-world complexity. We underestimate the friction.
The market is overpricing AI companies
OpenAI is chasing a $300B valuation while expecting $5B in losses on $3.7B revenue. AI companies aren’t SaaS. They don’t scale at zero marginal cost. Every prompt costs. Meanwhile, Microsoft and Meta are spending $300B this year on infrastructure just to keep up. The arms race is expensive and unsustainable.
The money isn’t in the models — it’s in the apps
AI models are becoming commodities. You can’t patent math. Open-source models are everywhere, and soon, they'll be on your phone, not locked in the cloud. The real edge? Building boring but brilliant apps — tools that transform HR, finance, compliance, and operations quietly but deeply.
Startups won’t win alone — incumbents have the edge
AI favors scale, capital, and data. Microsoft didn’t build the best video tool, but won by bundling Teams into Office 365. That’s the AI playbook. If you own the stack, pipelines, and trust — you don’t need the best model, just a good-enough one.
Generative AI isn’t the endgame
We’re obsessed with chatbots, but the future is Compound AI and Multimodal Systems — AI that can see, hear, analyze, and act in real time. Think less ChatGPT, more self-driving cognition. This demands rethinking data architecture, orchestration, governance, and resilience.
Can we think smartly about machines?
In 1950, Turing asked, “Can machines think?” Today’s better question: Can we think smartly about machines?
Integrate AI where it adds value
Rethink workflows, not just tools
Prioritize talent and architecture
Build for endurance, not demos
AI won’t be won by the loudest, but by deliberate, patient, focused leaders. Are you betting on models, or systems? Chasing hype, or building habits?
Comments