The AI Inflection Point Is Closer Than You Think

Earlier this month, Matt Shumer's essay "Something Big Is Happening" exploded across X, racking up over 80 million views and sparking debate about AI's accelerating capabilities and what it means for the future of work. We asked our Lead Software Architect, Vencel Lovasz, based in Budapest, to share his perspective. As someone who works with Claude models daily and watches the AI landscape evolve in real time, his take offers a grounded, thoughtful counterpoint from the engineering trenches:

These are genuinely difficult questions the author raises, and important ones. I've personally felt the pace of change accelerating. New models keep arriving, each one more capable than the last, able to solve more complex tasks with greater accuracy. And the release cycles are compressing. I've been a heavy user of Claude's Sonnet and Opus models, and looking back, Sonnet 4 came roughly a year after Sonnet 3.5, then Sonnet 4.5 followed just five months later, and now 4.6 arrived only four months after that. Each felt like a genuine leap, a new era in capability and accuracy. I believe this cadence will only continue to accelerate until we reach an inflection point: a moment of "high-level" maturity where gains begin to plateau.

The metaphor I keep coming back to, and I genuinely believe this maps closely to what's happening, is the development of the human brain. A child starts with no knowledge of the world and must learn everything from scratch. Each piece of knowledge builds on the last, and they gradually become smarter. That's where early AI models were. Then comes adolescence and early adulthood, when the brain develops at an extraordinary pace — near-exponential growth in intelligence and capability. That's where I think we are right now with AI.

This metaphor also holds for AI's expanding toolset. Just as a teenager and young adult gain new ways to interact with the world (a bank account, a phone, a driver's license, a computer), AI models are gaining more and more connectors and integrations through which they can act. The next phase, which Shumer also addresses, is AI reaching full "adulthood": the most capable, most reliable, most powerful models we've seen yet. When we arrive there, I think that will be the true inflection point. At that moment, the pace of AI evolution will slow. AI will remain deeply embedded in how we work and live, but the breakneck development cycle we're experiencing now will give way to something steadier.

As for whether we'll lose our jobs, honestly, I don't think so, and here's why: First, humans have an innate drive to find purpose. That's more philosophical than technical, but it's fundamentally how we're wired. We need something to live for, something to work toward. So we will always find something meaningful to do. Second, I don't see a fundamental difference between what's happening now and previous waves of industrial transformation. Before the industrial revolution, a person assembled a radio by hand; afterward, a machine did it. Before AI, a person wrote a blog post; going forward, a machine will generate one. Jobs will be displaced, yes, but they won't disappear. They'll transform.

This is already playing out in IT. Before AI, someone wrote a requirements specification or conducted a code review. Soon, a machine will handle much of that. But since AI lacks genuine judgment, a "soul," for lack of a better word, and I believe it always will, we will always need humans to supervise it: to check its outputs, craft effective prompts, and take responsibility for decisions. That human oversight role isn't going away.

There is one concern I do want to highlight, particularly in the software industry. If companies stop hiring junior developers — because AI can now do what a junior does, often better — then there will be no juniors to grow into mid-level and senior engineers over time. That talent pipeline is a real investment, and it's one that companies need to think carefully about. Hiring and mentoring junior developers isn't just charity; it's how the industry sustains itself. That, to me, is the most concrete and underappreciated risk in the current moment.