The first time I heard that testers would soon become unnecessary was around 2008, when Selenium RC started gaining popularity. The narrative back then was simple: everything will be automated. We will only need people who can write automation. Manual testing would disappear.Eighteen years later, QA is still here.
Today the conversation has shifted. With the AI boom, the focus is no longer on testers — it is on developers. We are told that programmers will soon be unnecessary because AI can write code. Everyone can now generate software. Within two or three years, most engineering jobs will disappear. After that, humanoid robots will replace the rest.
If that were true, QA would logically disappear even faster. If AI writes the code, surely it can test it too.
So I asked LinkedIn a simple question: When will AI replace most QA Engineers?
More than 3,000 people saw the poll. 233 voted. The results were interesting: 68% chose “Never.” Seven percent said within 10 years. Five percent said within 5 years. Ten percent said within 2 years. Three percent believed most QA would be replaced by 2026. I intentionally did not define what “Never” means. Predicting what happens in 1,000 years is pointless. I interpret “Never” as within our professional lifetime — the next 50 to 70 years.
Now let’s get to the core question.
Is it theoretically possible for AI to test a product completely autonomously? Let’s imagine we reach superintelligence, and it decides not to eliminate us. It builds products, writes code, and executes tests. Do we still need human testers?
If we build products for humans, then humans must ultimately evaluate their quality. It really is that simple.
AI can accelerate testing. It can generate test cases, simulate edge conditions, analyze logs, and improve coverage. It can automate enormous parts of the process. But quality is not only about assertions and passing checks. Quality is about usability, clarity, expectations, frustration, trust, and real-world experience.
If a system is built for another system, AI can evaluate AI. But if a product is built for people, human judgment remains essential.
And this leads to a bigger issue.
For the past four years, we have been told that AI will replace 95% of jobs within two or three years. This prediction has been repeated constantly since 2022. Every year, the same timeline: two or three more years. But four years have already passed.
If such a massive replacement was truly imminent, we would already see undeniable structural changes: entire professions collapsing, universities shutting down programs, mass technological unemployment driven directly by AI-native systems. Instead, what we mostly see is optimization.
Developers still develop — just faster. Designers still design — with generative assistance. Content creators still create — with AI drafts. Students still search for information — now through chat interfaces instead of Google.
The behavior model remains the same. The workflow is accelerated. That is optimization, not transformation.
When PayPal appeared, it changed how we send money. Gmail changed expectations about storage overnight. Facebook changed social interaction. YouTube changed media consumption. Shopify changed online commerce. The iPhone reshaped the entire mobile ecosystem. Netflix transformed distribution. Airbnb changed travel. Uber changed transportation behavior so deeply that children today cannot imagine calling a taxi by phone. Instagram reshaped visual sharing. Ethereum created a new financial layer. TikTok reinvented content discovery. Each of these products changed how people behave at scale.
Now it is 2026. ChatGPT launched in 2022 and triggered the modern AI wave. So here is the honest question: What product have we created with AI help in the last four years that has fundamentally changed human behavior at that level? None.
AI is clearly powerful. It is a horizontal layer across industries. It accelerates, enhances, and automates. But has it produced a new behavioral infrastructure comparable to the smartphone, ride-sharing, or algorithmic short-form media? So far, what we mostly see are optimizations built on top of existing models.
This does not mean AI is insignificant. It is extremely important. But there is a difference between a productivity multiplier and a civilization-level behavioral shift.
QA is not disappearing. The same applies to development. The stronger claim — that most jobs will vanish within two or three years — has been repeated for four years without materializing.
Maybe a true AI-native behavioral revolution is still ahead of us. But if the timeline is always “two or three years away,” then perhaps the prediction deserves more scrutiny than applause.
Daugiau apie testuotojo specialybę rašau savo knygoje https://itknyga.lt lietuvių kalba.
Top comments (0)