Introduction: Finding Echoes in History
As a trained economic historian specializing in 19th century's large technical systems like railways and telegraphs, I am always tempted to find historical parallels with today's emerging technologies—particularly what has come to be called artificial intelligence.
This impulse is not mere academic nostalgia. Understanding how past technological revolutions unfolded, who benefited from them, and why some innovations endured while others faded can offer crucial guidance for leaders, educators, and innovators navigating the current AI landscape. The question I keep returning to is simple but profound: Is AI genuinely transformative, or is it another overhyped technology destined to disappoint? How will we know?
Distinguishing Hype from Revolution
How do we know AI is not simply another hype cycle, like the Metaverse? One reliable indicator is academic engagement. When I examined publication trends, a striking pattern emerged. The Metaverse generated a brief surge of scholarly interest, but output declined sharply within a year. Researchers moved on because the underlying technology lacked depth, infrastructure, and practical application.
AI tells a different story. Academic publications on machine learning and large language models have grown exponentially for over a decade, showing no signs of decline. Researchers across disciplines, from computer science to medicine to education, continue to find substantive problems worth investigating. This sustained intellectual engagement suggests we are witnessing something more fundamental than speculative enthusiasm.
But academic interest alone does not guarantee transformative impact. What else distinguishes AI from previous hype cycles?
What Makes AI Different
Like the railways and telegraph, AI requires a large upfront investment in infrastructure Inevitably this creates over-investments, or a bubble where prices of assets are not related to any current earnings. Several structural factors, however, set AI apart from earlier technological waves.
First, AI builds in part upon existing mobile internet infrastructure. Unlike the Metaverse, which required users to adopt entirely new hardware such as goggles, and behaviors, AI integrates seamlessly into the mobile internet ecosystem already embedded in daily life. The smartphone in your pocket is already an AI delivery device. This dramatically lowers adoption barriers.
Second, the scale of investment is unprecedented. AI requires massive increases in energy consumption and computing power—what the industry calls "compute." Enormous data centers housing thousands of GPUs, each with a lifespan of just two to four years, represent billions in capital expenditure. The magnificent 7 drive Companies do not make such investments for passing fads. This infrastructure commitment signals serious long-term expectations.
The Magnificent Seven companies are a group of dominant U.S. tech stocks driving much of the S&P 500's performance: Alphabet (Google), Amazon, Apple, Meta Platforms (Facebook), Microsoft, Nvidia, and Tesla. Coined in 2023, these mega-caps (each over $1 trillion market cap as of late 2024) lead in AI, cloud computing, EVs, and social media, accounting for ~30% of the S&P 500's weight and outsized returns amid market concentration concerns. Although GDP is a flow variable and market capitalization is a stock variable, NVIDIA's marketcap is already 113% of Italy GDP in PPP terms, while Italy is 12th largest economy in the world.
Third, the application breadth of AI as a foundational technology is remarkable. Unlike single-purpose technologies, AI touches virtually every sector: healthcare, finance, education, manufacturing, creative industries, and beyond. This versatility suggests staying power. The adoption patern of the LLMs is also quite unique, organization struggle to integrate it due to inconsistent output and hallucination, but most professional individuals find it a useful time-saving writing and analysis tool for private use. These same individuals maybe relucatnat to use it professionally due to lack of guidance from their employers or fear of leackage of personal identifiable information.
The Divided Camps: Evangelists and Sceptics
Despite these indicators, opinion on AI's future remains sharply divided. In conversations at work and at social gatherings, I encounter two diametrically opposed camps.
The AI evangelists believe we stand at the threshold of a new era. They see AI solving our greatest challenges—accelerating scientific discovery, democratizing education, and creating unprecedented productivity gains. For them, skepticism is simply a failure of imagination.
The sceptics, however, urge caution. Some of AI's own founders have joined this camp, warning that we remain far from achieving artificial general intelligence. They point to the models' well-documented "hallucinations," inconsistent outputs, and tendency to produce confident nonsense. They predict that when the long-expected market correction arrives, most AI companies will vanish, leaving behind a graveyard of unfulfilled promises.
Both camps make valid points. The evangelists are right that the technology's potential is genuine and expanding. The sceptics are right that current limitations are real and that hype-driven valuations will eventually face reckoning. But I believe both camps miss something essential—something my two Italian great-grandfathers understood instinctively.
A Tale of Two Great Grandfathers
My great-grandfathers were born in the 1860s, a period of rapid and massive technological innovation. Railways were transforming commerce. The telegraph was revolutionizing communication. Steam power was reshaping industry. In 1900 the US patent office had issued so many patents that some believe nothing else could be invented. These men lived through changes as dramatic as anything we face today.
Odo Marcorini attended a technical marine college preparing him for a career in the commercial navy. Although he specialized in operating steam engines, he also learned Morse code. This relatively simple skill allowed him throughout his career to send and receive important messages independently. The telegraph was far removed from today's email, but the essentials—dates, times, costs, prices, and critical instructions—could be communicated over long distances at acceptable cost. Every railway line was flanked by a telegraph line, and the first undersea cables date from the 1850s. The Suez canal opened in 1867 was also flanked by a telegraph line. Knowing Morse code in those days was a badge of membership in a small, tech-savvy elite. After a career sailing around the worl that took him as far as China, Odo retired to a position at the Cassa di Risparmio in Verona, where he served until the end of his days.
My other great-grandfather, Bartolomeo Pighi, not only knew Morse code but built his entire career around the telegraph, serving as inspector of telegraph lines across the Province of Verona. His own father died young, leaving his formal education incomplete. But Bartolomeo or "Bortolo" compensated with limitless curiosity and willingness to work hard. His uncles were passionate nationalists who had fought alongside Garibaldi—one even at Bezzecca, the only Risorgimento battle the Italians actually won. Nonno Bortolo was a savvy investor and the first in the family to receive a regular salary. He was passionately interested in technology, purchasing an espresso machine and later a radio among the first in his community. Through his work as telegraph inspector, he played a key role in establishing post offices throughout Verona province in Santa Ana as well as Lonato del Garda, thus expanding the telegraph network. By the 1920s, he had accumulated enough to purchase a sixteenth-century villa with land producing olives and grapes. At the end of his life, he served as mayor of Montorio before passing away in 1939 just before the break-out of World War II.
What united these two men? They understood that emerging technologies reward those who invest the effort to learn them deeply. Morse code was the "prompting language" of their era—the skill that unlocked the telegraph's potential. Those who dismissed it as a passing novelty, did not understand the technology, or refused to learn its grammar were left behind.
Although Morse code for the telegraph was a deterministic system—the same input always produced the same output. Large language models are stochastic; identical prompts can yield different results. More critically, Morse code required learning a fixed grammar, while effective prompting remains an evolving, poorly understood skill even among experts. My point regards the willingness to learn a new "language" when a new technology emerges.
The Real Problem with AI Scepticism
This brings me to my central argument. The objections of some AI sceptics—particularly those who complain about hallucinations and poor-quality outputs—are often rooted in a failure to seriously learn how to use the technology. Other sceptics however are highly proficient users who have encountered real limitations: confidently wrong outputs in high-stakes domains, reproducibility problems in research, liability concerns in professional contexts.
If your prompts for the LLM's are unclear or unstructured, if you are using the wrong model for the task, what can you reasonably expect from the output? Complaining that AI produces nonsense when you have not invested in understanding how to communicate with it, is like my great-grandfathers' contemporaries dismissing the telegraph as unreliable because they never learned Morse code.
The parallel is instructive. Morse code required practice, precision, and an understanding of the medium's constraints. Effective prompting of large language models demands similar investment—clarity of thought, structured communication, and awareness of each model's capabilities and limitations.
Conclusion and Recommendations
So what should leaders, educators, and innovators take from this analysis?
First, recognize that AI is substantively different from previous hype cycles. The infrastructure investment, academic engagement, and breadth of application suggest genuine transformation, not speculative froth. This does not mean every AI company will survive or every promise will be fulfilled—but the underlying technology is here to stay.
Second, resist the temptation to join either extreme camp. Neither uncritical evangelism nor reflexive scepticism serves you well. Maintain clear-eyed assessment of both potential and limitations of a quickly developing technology.
Third, invest the effort to learn how to use AI effectively. Learning different prompt frameworks and structures is not as straightforward as learning to code, because even those who build these models do not fully understand how best to communicate with them. But that uncertainty is precisely why curiosity and experimentation matter. The rewards will flow to those willing to do the work.
Fourth, remember the lesson of my great-grandfathers. They belonged to a different age, but they understood something timeless: emerging technologies create opportunities for those who engage seriously with them. Morse code was their key to the telegraph revolution. Effective prompting may be yours to the AI revolution.
Put in the work. Stay curious. Experiment relentlessly.
The future belongs to those who learn.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.