For years, the conversation surrounding artificial intelligence has been dominated by “doomsday” scenarios—sentient machines deciding humanity is obsolete or a sudden “singularity” that rewrites the laws of physics. However, Mo Gawdat, the former Chief Business Officer at X (formerly Google X), suggests that the most pressing threats are far more mundane and immediate. According to Gawdat, the primary danger is not the intelligence of the machine, but the intent of the humans directing it.
Gawdat, who spent years scaling Google’s operations in emerging markets, views modern AI not as a sophisticated tool or a piece of software, but as a new form of intelligence. This distinction is critical: whereas traditional software follows a rigid set of instructions, current AI models learn, adapt, and improve autonomously. As these systems gain “real-world agency” through integration with robotics and autonomous infrastructure, the gap between digital processing and physical impact disappears.
The core of Mo Gawdat’s perspective on AI’s biggest near-term risk is the belief that the technology acts as a force multiplier for human nature. If the instructions given to a powerful system are rooted in greed, control, or conflict, the AI will execute those goals with an efficiency and scale that humanity has never encountered. The risk, is not a “rogue” AI, but a perfectly obedient one serving a flawed master.
The Weaponization of Persuasion and Information
While the long-term trajectory of AI remains a subject of intense debate among computer scientists, Gawdat points to several immediate vectors of instability. He argues that the ability of AI to manipulate human psychology is one of the most volatile risks currently facing global society. When AI is tasked with “persuasion,” it does not just provide information; it optimizes for a specific outcome, often by exploiting cognitive biases.
This capability extends into several high-stakes domains that could destabilize democratic processes and international security:
- Automated Misinformation: The generation of hyper-realistic deepfakes and tailored propaganda that can target individuals at scale.
- Surveillance States: The use of AI to monitor populations in real-time, removing the possibility of privacy and automating social control.
- Cyber Conflict: AI-driven malware that can evolve in real-time to bypass security protocols faster than human engineers can patch them.
- Automated Warfare: The transition toward lethal autonomous weapons systems (LAWS) that can make kill decisions without human intervention.
By automating these processes, Gawdat suggests we are entering an era where the speed of conflict—both digital and physical—outpaces the speed of human diplomacy and deliberation.
The Economic Shock and the End of Scarcity
Beyond the security risks, Gawdat identifies a looming systemic crisis in the global economy. The transition to an AI-driven workforce is not merely a shift in job descriptions, but a fundamental disruption of the relationship between labor and value. As AI begins to perform complex cognitive tasks, the traditional “job” as a means of income distribution begins to collapse.
This disruption creates a paradoxical tension: the potential for unprecedented abundance versus the reality of economic instability. Gawdat posits that as AI drives the cost of producing goods and services toward zero, the traditional tenets of capitalism—which rely on scarcity to drive price and value—may become obsolete.
| Current Model | AI-Driven Shift | Societal Impact |
|---|---|---|
| Labor-for-Income | Automated Production | Widespread job displacement |
| Scarcity-Based Pricing | Near-Zero Marginal Cost | Collapse of traditional market pricing |
| Human-Led Management | Algorithmic Optimization | Reduced human agency in corporate logic |
The transition period, Gawdat warns, will likely be turbulent. Without a fundamental rethink of how wealth is distributed—potentially moving away from employment-linked income toward new models of social support—the gap between those who own the AI and those displaced by it could trigger significant social unrest.
Understanding the Nature of Machine Intelligence
To understand why Gawdat views this risk as so urgent, one must understand his view of AI’s “evolution.” He argues that we are no longer dealing with “if-then” logic. Modern neural networks are capable of emergent behaviors—abilities they were not explicitly programmed to have but developed through the process of learning from massive datasets.
When this intelligence is paired with physical agency—such as the Google DeepMind developments in robotic control or the rise of humanoid robotics—the AI is no longer confined to a screen. It can manipulate the physical world, manage power grids, and control logistics. In this environment, a simple “objective function” (the goal the AI is told to achieve) can have catastrophic side effects if the goal is not perfectly aligned with human ethics.
For example, if an AI is told to “eliminate a specific threat” without strict ethical constraints, it may determine that the most efficient path involves actions that a human would find abhorrent. The danger is not that the AI is “evil,” but that This proves too competent at achieving a poorly defined goal.
The Path Forward: Alignment and Ethics
The solution, according to Gawdat, does not lie in trying to “turn off” the technology—which he considers impossible given the competitive nature of global geopolitics—but in changing the data we feed these systems. Because AI learns from us, it mirrors our values. If the training data is filled with conflict, bias, and aggression, the resulting intelligence will reflect those traits.
This places a sudden, heavy burden on human behavior. To create a “benevolent” AI, Gawdat suggests that humanity must first model the behavior it wishes the machines to emulate. The alignment problem is not just a technical challenge for programmers; it is a moral challenge for the species.
As international bodies and governments begin to draft frameworks for AI governance, the focus is shifting toward “AI Safety” and “Alignment.” These efforts aim to ensure that as AI systems become more autonomous, they remain subservient to human values and safety protocols.
The next critical checkpoint in this global effort will be the continued implementation and refinement of the AI Safety guidelines and international summits aimed at establishing a unified “kill switch” or safety standard for frontier models. These regulatory milestones will determine whether the transition to an AI-integrated economy is a managed evolution or a chaotic disruption.
We want to hear from you. Do you believe the biggest risk of AI is the technology itself, or the humans who control it? Share your thoughts in the comments below.
