Humanity is currently running the largest, most consequential experiment in its history. But unlike traditional scientific trials, this experiment isn’t confined to a sterile laboratory. It is happening in the open, connected to our global networks, and its subject is a technology that could soon outsmart us: Artificial Intelligence (AI).

As artificial intelligence rapidly evolves, a fear is taking root that AI will achieve a level of consciousness or independent autonomy that slips entirely beyond human control. If that happens, we will have created a technologically “organic” being in the digital realm. This wouldn’t just be a tool that complements our workflows; it would be an entity with the capabilities to replace and outmanoeuvre us, potentially turn the tables against its creators.

The world as a beta test

When a nation develops a new weapons system, whether a hypersonic missile or a biological agent, the testing takes place in highly secure, classified facilities. It happens in barren deserts, isolated underground bunkers, or Biosafety Level 4 laboratories. The environment is rigidly controlled to contain the blast radius or prevent a catastrophic leak.

AI, arguably a more unpredictable force, is being tested directly on us. This is what tech ethicists mean by the “open-air lab.” By releasing advanced AI models to the public before fully understanding their safety profiles, tech companies have turned the entire world into a testing ground. Every time you type a prompt into a chatbot, ask an image generator to design a logo, or correct an automated transcription, you are unwittingly participating in this mass experiment.

We are training the AI to be smarter and more capable, yet the long-term consequences are treated as an afterthought. We are simultaneously the lab assistants and the guinea pigs. The tech industry has deployed these systems with a “release first, patch safety later” mentality, making this open-air experiment arguably more dangerous than conventional weapons testing, which at least respects the boundaries of a controlled environment.

Three reactions to the unknown

Basically, humanity has engineered a technology it does not fully understand and we are fundamentally divided on how this will affect our future into the following categories:

• The accelerationists: This group believes we should let AI grow unhindered until it reaches superintelligence, hoping it will cure diseases and solve climate change.
• The regulators: This faction argues for international coalitions and guardrails to ensure AI models align with human values before they are integrated into critical infrastructure.
• The doomers: This group seeks to halt advanced AI development entirely, viewing superintelligence as a pandora’s box that cannot be closed.

The implications

Because this mass experiment is taking place in an open world, any negative consequences cannot simply be retracted. Currently, experts are tracking two major implications of this unchecked trajectory and both carry dire consequences for humanity.

Implication 1: social upheaval

Before AI ever poses a physical threat, it is already rewiring the fabric of society. The consequences here are not just theoretical; they are already proving dire for humanity, eroding our economic stability, truth, and fundamental rights.

We are seeing real-world damage unfold in real time:

• Job displacement and labour strikes: The recent historic strikes in Hollywood were largely driven by writers and actors fighting for their livelihoods against studios planning to use AI to generate scripts and create perpetual digital replicas of performers. Entire freelance markets in copywriting, translation, and illustration have already collapsed.
• The death of truth: AI-generated deepfakes have been weaponised with devastating effect. We have seen AI-generated audio used to impersonate political figures in attempts to suppress voter turnout, and malicious actors using AI to create non-consensual explicit imagery of private citizens and public figures alike, ruining reputations irreparably.
• Algorithmic bias: Real-world AI screening tools used for hiring, lending, and law enforcement have repeatedly been caught scaling human prejudices—denying marginalised groups housing loans or falsely flagging individuals for crimes they did not commit.
• AI reliance: As humans increasingly rely on AI for decision-making and information processing, our cognitive habits are changing. We are outsourcing our critical thinking to algorithms, fundamentally degrading what it means to be human.

Implication 2: Technological destruction

While the social upheaval is dire, the second implication points to a different kind of dire consequence for humanity: physical and existential destruction. The warnings are coming directly from the architects of the technology itself. Geoffrey Hinton, the Nobel Prize-winning computer scientist widely known as the “Godfather of AI,” recently warned that humans may have “maybe a year, but not much more than a year” to establish safeguards before AI becomes incredibly difficult to control. His primary concern is “recursive self-improvement”, a scenario where AI systems become capable of designing and improving their own successors without human involvement, rapidly outpacing our ability to pull the plug.

Similarly, Dario Amodei, the CEO of the frontier AI company Anthropic, has publicly called for the industry to pace itself. Amodei has warned that building AI too fast is “reckless,” urging independent evaluators and safety guardrails. When the very creators of technology, the people who stand to profit the most from it, are asking for the brakes to be pulled, society must pay attention.

Time to return to the lab: Reclaiming the ethics of innovation

We are standing at a critical juncture. We have allowed technology with the potential to supersede human intelligence to be beta-tested on the public in an uncontained, open-air experiment. To fix this trajectory, humanity must urgently return to the fundamental ethics of technological invention.

Throughout history, responsible innovation in medicine, aviation, structural engineering, and biotechnology has been governed by a sacred principle: the precautionary principle.

Before any invention with high potential harm is introduced to the public, the inventor carries an ethical obligation to prove it is safe in isolated, controlled environments.

The rule was clear: first, do no harm and never expose society to unquantified risks for commercial speed.

AI development has completely inverted this historical ethical contract. The tech industry’s “move fast and break things” ethos replaced rigorous containment with real-world exposure, forcing society to absorb the collateral damage as an unintended consequence.

It is time to reinstate those foundational ethical boundaries. We must treat frontier AI not as a consumer software update, but as a high-stakes technology requiring strict containment. Returning to ethical invention means:

• Enforcing the precautionary principle: Requiring AI developers to unequivocally prove that a model is safe, controllable, and aligned with human values before it is deployed.
• Ending the open-air lab: Pulling autonomous, ultra-capable models back into isolated sandbox environments where risks can be studied without risking global infrastructure.
• Human-centric accountability: Refusing to let commercial races override human rights, economic stability and long-term existential safety.

If we are building an intelligence that could one day surpass us, returning to the ethics of creation isn’t just an academic preference. It is our only remaining safeguard.

(Chin Yee Mun is Director, Centre for Social and Policy Studies, Tun Tan Cheng Lock Institute of Social Studies, UTAR.)