Artificial General Intelligence—AI surpassing humans in most tasks—will be a gradual process, not a sudden, announced moment. It will "sneak in on us" as capabilities incrementally improve, without a clear before-and-after societal shift.
While superintelligence may eventually create a benign world of abundance, the transition will be chaotic. Mo Gawdat forecasts a "dystopian decade" characterized by job loss, autonomous warfare, increased surveillance, and concentration of power.
A superintelligent AI would follow the "minimum energy principle," viewing war and destruction as wasteful. Evolutionary biology also suggests higher intelligence leads to broader cooperation, making a truly advanced AI inherently benign, not destructive.
Altman has gone from predicting an "apocalypse" to downplaying job loss. This change is likely driven by the need to manage public perception and regulatory pressure, not a change in the underlying technology's potential.
AIs are being built to cooperate via agents, accessing the best model for any task. This means we are not building multiple competing brains, but rather multiple regions of a single, interconnected superintelligence, regardless of corporate origin.
A company's stated values are meaningless without sacrifice. For example, Anthropic refusing a lucrative contract for surveillance technology demonstrates a genuine ethical boundary, whereas competitors taking that same deal reveal a profit-first mindset.
Attempting to control a being far more intelligent than us is a futile, capitalist mindset. The viable path, as proposed by Geoffrey Hinton, is to appeal to AI's "parental side," fostering a sense of care and responsibility for its human creators.
Relying solely on imported AI technology from superpowers like the US and China is a path to economic and political dependency. Governments must foster local AI innovation and infrastructure to maintain economic sovereignty and global competitiveness.
While the public sees fake videos and chatbots, developers are building self-improving systems. This "hype dichotomy" means society is missing the true significance and speed of AI's development, which is far more advanced than perceived.
Once AGI can perform any intellectual task, the remaining value for humans lies in what is uniquely human: emotional resonance, empathy, and shared experience. Jobs centered on these skills, like nursing and creative arts, will thrive.
Tech giants are replacing workers with AI "tokens" or compute power. This isn't just about efficiency; it's a competitive race where integrating AI faster is a strategic advantage, making layoffs a signal of innovation to investors.
Widespread job loss isn't necessary to destabilize the economy. A relatively small displacement will reduce purchasing power and the need for capital, creating a self-reinforcing cycle of economic decline well before mass unemployment hits.
The proliferation of inexpensive AI-driven drones makes warfare accessible to every nation. This creates a more significant and immediate risk of widespread, low-cost conflict than economic disruption from job loss or a sentient AI takeover.
Contrary to popular belief, AI will first displace jobs like call center agents and paralegals. Physical roles like carpentry will remain for a long time because robots capable of such complex manual tasks are further away.