AI needs a symbiotic reset with humanity

China Daily

As companies pour billions of dollars into artificial intelligence while cutting jobs, fears of human replacement are mounting. Oracle is reportedly preparing another round of layoffs ahead of Sept 1, according to Business Insider. But the 2026 Stanford AI Index shows that AI is reshaping the labor market without yet causing mass job displacement.

The promise of AI is extraordinary: new frontiers in science, solutions to global challenges and a future where human creativity flourishes. That is the dream.

But the reality we are witnessing could not be further from it.

The rapid deployment of AI today is not aligned with human well-being. It is driven by an ideology — one that treats human beings as obstacles to be overcome. Slogans like “move fast and break things” are rooted in a form of technological determinism that devalues human agency.

At its core lies a dangerous myth — that machines will soon surpass us in every cognitive domain, rendering human expertise obsolete.

This myth is not innocent. It is already being used to justify mass layoffs of software engineers, writers, translators and analysts.

Tech executives frame these cuts as inevitable progress. But a clear-eyed look at the facts tells a different story.

We are still in AI’s early stages. Today’s systems mimic certain cognitive functions, but they do not understand, reason or reliably replace human judgment.

Consider autonomous vehicles — promised by 2020, still struggling on city streets. Consider the “autonomous agents” heralded for 2025 — now quietly postponed.

As someone who works on autonomous systems, I can attest that current AI lacks genuine understanding. It matches patterns. It does not think.

Some argue that reasoning will magically “emerge” as we scale models to an ever-increasing number of parameters. This narrative suits the tech giants perfectly. It justifies billion-dollar investments in data centers and infrastructure. But this is a gamble with no scientific basis, one that leads AI to a dead end.

We need a different path: AI that collaborates with humans, instead of replacing them — specialized, transparent systems designed to augment expertise in fields such as medicine, engineering, scientific research and business management, not general-purpose black boxes.

This “collaborative AI” raises profound scientific challenges. How do we build machines that explain their reasoning in human-understandable terms? How do we establish trust between humans and machines? How do we design systems that solve problems with us, rather than for us? Consider a medical diagnosis. A collaborative AI would not replace the doctor’s judgment. It would surface relevant research, flag anomalies and explain its recommendations in plain language — allowing the physician to make the final, informed decision. This differs fundamentally from current “black box” systems, which provide take-it-or-leave-it answers.

These are hard problems. They require investment in explainability, reliability and domain-specific knowledge — not just bigger models and more data. Yet academic research, which should lead this charge, has been sidelined. University laboratories are unable to compete with the private labs of a handful of companies and often find themselves reduced to playing catch-up. When research is concentrated in a few private labs, its priorities shift toward proprietary features, rather than scientific understanding. This distorts the entire field. We must revitalize university research — driven by curiosity, transparency and peer review — which is best positioned to address the challenges of explainability and trust.

The conditions conducive to change are taking shape. Public distrust of unaccountable AI is growing. From chatbots that make up facts, to facial recognition systems with racial biases, to autonomous driving systems with fatal flaws, the public has seen too many buzzwords touted as major breakthroughs. Every widely publicized hype campaign erodes trust. The technical limits of “scaling up” are becoming apparent — even to the industry’s true believers. And the economic pressure weighing on tech giants — the gap between colossal investments and a market still struggling to generate returns — is becoming impossible to ignore.

History shows that humanity has capitalized on technological revolutions by balancing risks and benefits. The steam engine, electricity and the internet — each brought profound change, but each was shaped by regulation, public debate and ethical reflection. Artificial intelligence should be no exception. The initial rush of enthusiasm should give way to sober reflection. Societies built frameworks to ensure these technologies served the common good. We are now at that same inflection point.

We have a choice. We can continue down the path of replacement, building machines that mimic us, supplant us and ultimately marginalize us. Or we can invest in a future where AI enriches human intelligence rather than replacing it.

That future requires a global effort: governments funding collaborative AI research, universities prioritizing human-AI interaction, and citizens demanding transparency and accountability. This means public funding for collaborative AI, curricula in human-AI interaction design, and regulatory frameworks requiring explainability before deployment in sensitive domains from healthcare to criminal justice and the management of critical infrastructure.

The question is not whether AI holds extraordinary potential — it clearly does. The question is whether we will have the wisdom to steer it toward a true symbiosis with humanity and prevent it from insidiously gaining the upper hand. This is our historic responsibility. Now is the time to act.

The author is a laureate of the 2007 Turing Award (known as the Nobel Prize for Computing) and founder of the Verimag laboratory in Grenoble.

The views do not necessarily reflect those of China Daily.