True transformation takes time

A recent article from Bloomberg—AI Will Upend a Basic Assumption About How Companies Are Organized by Azeem Azhar—is like many in a long string of articles that have pumped up the disruption and transformation being brought on by artificial intelligence. Azhar also wrote the 2021 Financial Times Best Book of the Year, The Exponential Age. This view contrasts with other news, for example, a recent article from Gizmodo, Analysts Notice Microsoft Quietly Cancelling Data Center Leases by Matthew Gault. The article says, "After promising to spend $80 billion on AI infrastructures, the tech giant appears to be pulling back." Microsoft CEO Satya Nadella recently said, "The world has yet to turn any of today’s AI hype and spending into a meaningful lift in the actual economy."

This tension between hype and reality encapsulates where we are with AI today. On one hand, there’s no doubt that tools like ChatGPT, Claude, and others are getting better. On the other hand, most large language models (LLMs) still have a problem with facts. LLMs seem to have a desire to please their overlords because sometimes they will invent papers and cite articles that don't exist. This was a problem early on but remains an issue even today. Perhaps because these models are trained on content people create—and people have recently taken to denying basic facts—they can't help themselves.

As someone who uses AI nearly every day and manages an AI team at work, I see its potential firsthand. But I also see its limitations. On the ground, Nadella’s view resonates: companies aren’t seeing AI money rain from the sky just yet. And while we’re clearly in the early days of interactive AI, I can’t help but think back to the early days of the web in 1994. It opened up a new world of information but took nearly a decade before it really hit the mainstream. Now that we are 30 years from the birth of the web, it’s clear that the web transformed the global economy and politics. I suspect that AI will go on to do the same—in time. For now, though, I would ignore the hyperbolic claims that seem to be coming mostly from academics and media commentators.

The Dramatic Impacts of AI

While Nadella’s tempered view resonates with many in business today, there are undeniable areas where AI has already had a profound impact.

• Business Transformation: In industries like healthcare, logistics, and finance, AI is fundamentally changing how companies operate. Predictive analytics tools powered by AI are enabling hospitals to anticipate patient needs and optimize staffing levels. Similarly, supply chain management has been revolutionized by AI-powered platforms that can forecast demand or reroute shipments based on real-time disruptions.
• Market Disruption: One striking example of AI’s disruptive potential came earlier this year when DeepSeek released its R1 open-source AI model. The model’s efficiency challenged Nvidia’s dominance in high-performance computing hardware for AI applications, leading to a staggering $560 billion loss in Nvidia’s market value in just one day. Events like this highlight how advancements in AI can destabilize even the most entrenched players in an industry.
• Generative Creativity: Generative AI tools like ChatGPT or MidJourney are reshaping creative industries by automating tasks such as graphic design or content creation. Companies can now produce personalized marketing campaigns or product prototypes at scale—a capability that was unthinkable just five years ago.
• Healthcare Innovation: In medicine, AI is accelerating drug discovery by analyzing molecular interactions at speeds impossible for humans. Tools like AlphaFold have already revolutionized protein structure prediction, opening doors for groundbreaking treatments.

These examples align with what Azeem Azhar describes as “exponential” change—where technology accelerates faster than organizations or societies can adapt. However, as Nadella points out, these disruptions haven’t yet translated into broad economic gains.

The Subtle Yet Pervasive Impacts

Not all changes brought about by AI are headline-grabbing; many are quietly improving efficiency and convenience in our daily lives.

• Everyday Interactions: From personalized Netflix recommendations to Google Maps’ real-time traffic updates, AI has become an invisible force shaping our routines. Smart home devices like Alexa or Nest thermostats use machine learning to adapt to user behavior without fanfare.
• Workplace Efficiency: In offices worldwide, AI is embedded into tools like Microsoft Excel or Salesforce dashboards to provide actionable insights without requiring users to learn new systems. For example, sales teams now rely on predictive algorithms to identify high-potential leads based on historical data.
• Consumer Personalization: Companies across industries are using sentiment analysis tools to adjust their messaging in real time based on customer feedback. Retailers like Amazon employ machine learning models to personalize shopping experiences down to individual preferences.

Here’s where my own experience comes into play: while journalists and academics often focus on grand narratives about disruption, much of what I see day-to-day is incremental improvement rather than revolution. For instance, small-to-midsize businesses (SMBs) are increasingly leveraging affordable AI tools for tasks like automating customer support or managing inventory. Platforms like Shopify integrate machine learning algorithms that recommend pricing strategies or predict seasonal demand spikes.

These quieter impacts may not grab headlines—but they’re reshaping how we live and work in ways that feel almost invisible until you take a step back.

Balancing Hype with Reality

Despite these advancements, it’s crucial to separate hype from reality when discussing AI’s economic impact. Economist Daron Acemoglu estimates that over the next decade, AI-driven productivity will contribute only a modest 0.7% increase in GDP growth for the U.S., largely due to implementation costs and organizational inertia offsetting gains.

Even within companies actively adopting AI, challenges remain significant. Hallucinations—where large language models generate false but plausible-sounding information—continue to undermine trust in these systems. This issue reflects broader concerns about data quality and algorithmic bias as businesses scale their use of AI technologies.

As I mentioned earlier, LLMs often "hallucinate" or fabricate information because they’re trained on imperfect human data—and this problem hasn’t gone away despite advancements in model accuracy.

Looking Ahead

The trajectory of artificial intelligence mirrors that of past technological revolutions like the internet—it promises transformative change but requires time for infrastructure and adoption to catch up with innovation.

Over the next few years:

• Industries such as pharmaceuticals and telecommunications are expected to lead in adopting vertical-specific solutions tailored for their unique challenges.
• Edge computing will enable smaller yet powerful models capable of running locally on personal devices, reducing dependency on cloud infrastructure.
• Regulatory frameworks around data privacy and ethical use will shape how quickly businesses can deploy advanced systems at scale.

For now, Nadella’s skepticism serves as a useful counterbalance to media narratives touting overnight revolutions driven by generative models like ChatGPT or Claude. As someone who uses these tools daily—and sees both their potential and limitations—I believe we’re still figuring out how best to integrate them into our lives meaningfully.

The genie may indeed be out of the bottle—but as history shows us with innovations like electricity or the internet—true transformation takes time.