Photo Courtesy of EXANTE
The feeling is remarkably familiar, carrying a distinct scent that seasoned technologists recognize instantly. It is a heady mix of frantic market enthusiasm, inflated financial projections, and the pervasive anxiety that entire professions are about to be swept away by a technological tidal wave. For Richard Forss, Chief Technology Officer at global multi-asset prime broker EXANTE, this atmosphere is not entirely new. It stirs up memories of a different era—one defined by dial-up tones, high-stakes financial bets, and a very similar brand of corporate panic.
Back in 2000, Forss was working at Deutsche Bank, tasked with building a brand-new trading platform. As he sat at his desk, he watched his friends and colleagues pack up their belongings one by one, leaving traditional banking behind to join high-flying internet startups that promised quick fortunes. They were armed with stock options and visions of early retirement, mapping out leisurely lives by the time they reached forty. Meanwhile, Forss stayed at the bank, grappling with the quiet envy of a man performing sensible, grounded work while everyone around him seemed to be participating in a massive, lucrative windfall.
Then the dot-com crash arrived, and the supposed goldmine evaporated almost overnight. Companies that had commanded massive valuations on the back of little more than a domain name and a business plan vanished from the ledger. Pets.com famously went from a high-profile Super Bowl advertisement in January to complete liquidation by November of the same year. The Nasdaq lost the vast majority of its value, wiping out trillions of dollars in market capitalization. Thousands of fledgling enterprises died in the fallout, but the web itself did not perish. It quietly carried on, laying the foundational infrastructure that eventually built giants like Google. Forss was still at his desk at the bank, and looking back, he has rarely felt so grateful for a boring, conservative decision.
Forss mentions this foundational career experience because he sees the exact same psychological climate gripping the tech industry today, and he recognizes the underlying mechanics at play.
The Oldest Trick in the Industry
To understand the current frenzy surrounding artificial intelligence, it helps to look past the marketing departments and examine the linguistic roots of the term itself. "Artificial intelligence" is, at its core, a marketing construct. It was coined in the 1950s by a group of computer scientists who needed a compelling hook to secure funding from the United States military. The strategy worked brilliantly—which remains the primary reason the phrase is still in active circulation today.
Every decade or so, this umbrella phrase gets dusted off, heavily oversold to the public and investors alike, and invariably followed by a period known as an "AI winter," during which venture capital and institutional funding dry up. Depending on how one counts the historical iterations, the technology sector is currently riding the fourth or fifth turn of that recurring boom-and-bust wheel.
The underlying technology itself is undeniably real, of course. Deep learning achieved a startling breakthrough at the ImageNet computer vision competition in 2012, demonstrating capabilities that caught even computer scientists off guard. Within three years of that milestone, an ordinary smartphone in a consumer’s pocket possessed the capability to accurately identify the contents of a photograph. However, real technical capability and honest, measured expectations are two entirely different things, and the vast gap between them is precisely where speculative fortunes are made and lost.
The historical pattern never seems to change. A genuine, hard-fought breakthrough arrives in a laboratory, and institutional capital floods in behind it to capture the upside. The promises made by executives and founders inflate rapidly to justify skyrocketing market valuations, eventually reaching a fever pitch where someone stands on a stage and declares that half of all corporate engineers will be rendered redundant by Tuesday afternoon. Then, reality inevitably sets in, the speculative money retreats, and the engineers and practitioners who are still standing simply roll up their sleeves and go back to work.
The Radiologists Who Never Disappeared
A striking historical parallel to today’s panic over AI displacing software engineers can be found in the medical field nearly a decade ago. In 2016, Geoffrey Hinton—widely recognized as one of the founding godfathers of deep learning technology—stood before a packed room and advised medical institutions to stop training human radiologists altogether. His reasoning was that advanced deep learning models would outperform human specialists in image analysis within five years, making the profession obsolete.
As it turned out, that prediction was fundamentally mistaken. The advent of cheaper, faster medical imaging did not reduce the need for human experts; instead, because scans became quicker and more affordable to produce, hospitals and doctors ordered them in vastly greater quantities. Consequently, the overall demand for radiologists actually climbed rather than plummeted. When a specific task becomes dramatically cheaper and faster to execute, human society almost universally responds by wanting more of it, not less.
Forss reflects on that healthcare lesson every time someone suggests that his own profession is nearing its end. The latest wave of automation tools has firmly landed in the software development sector. By 2025, agentic coding assistants ceased to be mere party tricks or novelty demonstrations and began closing real, production-grade work. These modern systems can take a software issue, write the necessary code, and open a pull request that a human developer can then review, debate, and refine. Engineers at firms like EXANTE use these tools daily, and they ship software products faster as a direct result.
Yet, what has distinctly not happened is the wholesale replacement of the human workforce. The advanced tool still requires a responsible adult in the room—someone capable of distinguishing a plausible-sounding answer from a correct one, and someone who is on call when the system inevitably runs into a critical failure at 9:31 on a Monday morning. That indispensable adult is the software engineer. In practice, there is simply a higher level of demand now for the complex problems that organizations can finally afford to solve.
The Point We Keep Forgetting
A prime example of how these technologies truly drive progress can be seen outside the realm of commercial software development. DeepMind did not rely on conventional, hand-written software engineering to crack the complex biological problem of protein folding—a monumental scientific hurdle that had systematically defeated human biologists for half a century. Instead, the organization built an advanced machine learning model that successfully solved the problem, and then generously published the structural maps of 200 million proteins for free use by the global scientific community.
Real-world benefits materialized almost immediately. Groundbreaking research into new antibiotics and advanced malaria vaccines has already emerged directly from that single, open-access act. When historians look back at that achievement, nobody asks how many lines of traditional code it took to reach the finish line.
That remains the fundamental lesson buried beneath the current industry noise. Software developers were never hired purely to write lines of syntax; the code was always merely the means to an end. The actual job of the engineer was—and continues to be—solving complex problems for human beings.
After spending more than twenty years occasionally mistaking the tool for the mission, practitioners are finding that the emergence of advanced machines is serving as a welcome reminder of what engineering was always supposed to be about.
Richard Forss is CTO of EXANTE, a global multi-asset prime broker, where he leads a 230-person technology division—none of whom he has yet worked out how to replace with a machine, himself included.
Comments (0)