AI: The Revolution That Looks Suspiciously Like the Last One
It has become almost impossible to scroll through a newsfeed without stumbling over a freshly baked opinion on artificial intelligence - ranging somewhere between “just another shiny gadget” and “end of civilisation as we know it.” The drama is palpable. Reading some of these takes, I often wonder whether we haven’t reached the point where one would need a “reverse Turing test” just to spot the humans amidst the hallucinations. In other words, it’s increasingly hard to tell whether the feverish commentary on AI comes from an actual person, or from the very chatbot they’re supposedly warning us about.
My own perspective is considerably less operatic. AI is, in essence, just another incremental step in a very long sequence of incremental steps. Typewriters gave way to electric typewriters, which in turn nodded politely and shuffled off stage when computers with keyboards took over. Each wave brought the usual benefits - efficiency, convenience, cost savings - and the usual new headaches: viruses, software upgrades, and the eternal punishment of using Windows in the workplace. From a labour perspective, the cycle is equally predictable - repetitive entry-level tasks disappear, work requires more skills, and humans reluctantly specialise. Déjà vu, but with shinier fonts.
If you zoom out, the conversation around displacement of junior coders is not a brand-new tragedy either. We already had auto-complete tools, a Stack Overflow full of “ready-made” answers, reusable component libraries, drag-and-drop website builders, and visual programming. To bemoan that AI tools might accelerate this existing trend is a little like protesting that power steering has ruined the art of driving: the ship sailed long ago.
Will AI accelerate things? Almost certainly. Just as computer chips are designed with the help of earlier computer chips, AI will inevitably be used to design, test, and refine the next generation of AI. The layering of abstractions - libraries upon libraries, agents upon agents - is well underway. The complexity is such that no individual human can fully grasp the systems they’re building, which makes it unsurprising that we would outsource parts of their creation to, well, systems that don’t complain about overtime or coffee budgets.
And will that “change the world”? Probably - but in smaller, more uneven ways than the prophets and doomers imagine. Remember robotic process automation? It was once hailed as a revolution, yet MIT reports that 95% of AI projects lead nowhere. On the other hand, we should recall that the internet, too, was once written off as a fad or reduced to a “faster fax machine.”
New tools are always used first to imitate the old. Only much later do their genuinely transformative uses emerge. At our consulting company, we see AI as exactly that: not a doomsday engine, but a pencil. And, as any designer knows, a pencil in an amateur’s hand makes stick figures; in the hands of an expert, it produces depth, shading, imagination. We’re still in the stick-figure phase of AI. The real applications - layered through software, embedded in objects, roaming in robots - are still ahead of us. The show has barely started.
Stefan Dulman