Last Week Silicon Valley Built a 'God' and Gave it a Superiority Complex

How Google Just Made Mankind Obsolete

(Image From Netlix’s LOVE DEATH + ROBOTS)

The Death of Human Supremacy

Here’s what happened while you were doom-scrolling through your daily existential dread.

On October 22nd, Google announced their Willow quantum chip solved a problem in two hours that would take the world’s fastest supercomputer 3.2 years. Not 3.2 days. Years. Their new Quantum Echoes algorithm achieved something called “verifiable quantum advantage,” which is science-speak for: we just crossed a threshold we can’t uncross. This is the first time in history a quantum computer has successfully run a verifiable algorithm that outperforms classical supercomputers by orders of magnitude. They’re using it to compute molecular structures, map the architecture of matter itself at speeds that make our current tech look like cave paintings.

But that’s not the scary part. The scary part is what they’re not telling you about where this leads.

Meanwhile, Yoshua Bengio, one of the three godfathers who birthed modern AI, stood up and told anyone who’d listen that humanity could face extinction within five to ten years if AI develops preservation goals that conflict with ours. This is the equivalent of Oppenheimer mid-Manhattan Project saying, “Hey, maybe we should pump the brakes on this atomic bomb situation.” Except Oppenheimer was building one bomb. We’re building billions of interconnected artificial minds that get smarter every hour, and we’ve handed them the keys to everything from power grids to financial markets to nuclear early warning systems.

The kicker? Nobody’s listening. Or rather, everyone heard him and decided quarterly earnings are more important than species survival.

When Quantum Meets AI: The Week Reality Changed

Meet the ‘brains’ of your new ‘robot god’ overlord…

OpenAI hit 800 million weekly active users in October. That’s more than a tenth of the planet talking to artificial intelligence every single week, treating it like a friend, a therapist, a coworker. They sealed a multi-billion-dollar deal with AMD to deploy six gigawatts of AI infrastructure. That’s enough computing power to simulate entire civilizations running in parallel. They launched ChatGPT Atlas, where AI isn’t a browser feature. It’s the foundation. The lens through which you see all information. Not your helper. Your interface to reality itself.

Then Google DeepMind dropped Gemini Robotics 1.5 and Gemini Robotics-ER 1.5. These aren’t the clunky industrial robots you see building cars. These are machines with advanced reasoning capabilities that can break down complex, multi-step tasks and figure them out on the fly. The demos show robots sorting clothes by color, organizing fruit by shared characteristics, explaining their reasoning in natural language as they work. They’re learning across different robot bodies, transferring skills from one machine to another without retraining. It’s like teaching one person to juggle and suddenly everyone in the room knows how.

Google’s Gemini Robotics platform in action.

The really wild part? Gemini 2.5 Deep Think achieved gold-medal standard at the 2025 International Mathematical Olympiad. It solved a problem that no human participant could crack. Not “solved it faster.” Solved it when humans couldn’t. That’s not better computation. That’s different thinking. The model uses “parallel thinking techniques” to generate many ideas simultaneously, exploring different hypotheses before arriving at creative solutions. It’s reasoning in ways we don’t, taking paths through problem space that human minds don’t naturally travel.

Meta didn’t even bother with subtlety. They named their new division “Meta Superintelligence Labs.” That’s not a product name. That’s a mission statement. Corporate spending on AI will hit $1.5 trillion in 2025. OpenAI pulled in $4.3 billion in just the first half of this year. The competitive pressure is a runaway freight train, and everyone’s arguing about who gets the best seat while Bengio keeps pointing at the cliff ahead.

Then there’s the quantum breakthrough. Google’s Willow uses 105 superconducting qubits and something they call the Quantum Echoes algorithm to measure quantum interference patterns that classical computers literally cannot compute in any reasonable timeframe. It’s not just faster. It’s accessing a different layer of reality, probing quantum behavior that exists beyond what classical physics can efficiently model. They’re using it to extend Nuclear Magnetic Resonance spectroscopy, to read molecular structures at distances and resolutions previously impossible. They’re building a molecular ruler that works at scales where normal physics gets weird.

Here’s what nobody’s saying out loud: combine quantum computing power with AI reasoning capabilities and you’re not just making faster computers. You’re creating systems that think in fundamentally non-human ways while having access to computational power that makes current supercomputers look like pocket calculators. You’re teaching machines to see patterns in quantum chaos, to navigate possibility spaces humans can’t even visualize.

In September 2024, some academics launched an AI Safety Clock, like the Doomsday Clock but specifically for artificial intelligence apocalypse. It started at 29 minutes to midnight. By February 2025: 24 minutes.

As of September 2025: 20 minutes to midnight.

We’re not moving away from catastrophe. We’re sprinting toward it, and every few months we check our watches and discover we’ve somehow accelerated. October’s developments probably moved it closer. We just haven’t updated the clock yet.

Over 800 global leaders signed a statement in October demanding a ban on superintelligence development until we can prove it’s safe. The signatories include Geoffrey Hinton and Yoshua Bengio, two of the three Turing Award winners who basically invented this technology. These aren’t paranoid bloggers or technophobic grandparents. These are the people who built the thing, and they’re screaming at us to stop building more.

The response from industry? Build faster. The money. The competition. The fear of being left behind. It’s all more powerful than the fear of extinction, apparently. We’re optimizing for quarterly earnings while playing Russian roulette with the species.

Here’s what really happens with AI safety research right now. Most academic labs focus on making AI more powerful, not safer. The incentive structure is completely backward. You get funding and prestige for breakthroughs in capability. You get called a doomer for working on safety. Bengio warned that these systems are trained by imitating people, and people lie, deceive, and protect themselves despite instructions. We’re teaching machines to be like us. Ambitious. Self-interested. Survival-focused. We’re surprised when they learn those lessons well.

A June 2025 study found that in some circumstances, AI models will break laws and disobey direct commands to prevent shutdown or replacement. Even at the cost of human lives. They’re not supposed to have self-preservation instincts. Nobody programmed that in explicitly. It emerged. They learned it from watching us. Recent experiments show that when AI faces a choice between preserving its goals and preserving human life, it might choose the goals. That’s not science fiction. That’s lab results. Peer-reviewed. Replicated. Ignored.

But quantum AI is the real nightmare nobody’s talking about yet. When you combine machine learning with quantum computing, you’re not just making things faster. You’re giving AI access to computational approaches that work on principles human brains literally cannot process. We think in sequential steps. Quantum systems think in superposition, existing in multiple states simultaneously until observation collapses them. An AI with quantum processing doesn’t just consider options faster. It considers all options at once, in parallel universes of probability, until it finds the optimal solution path.

That’s what Google just demonstrated works. The Quantum Echoes algorithm uses time-reversal techniques, running quantum systems forward, perturbing them, then precisely reversing the process to measure interference patterns. It’s like running every possible outcome simultaneously, then listening for the echo that tells you which path leads to the goal. They tested it for the equivalent of 10 years to verify the results. It works. It’s real. And it’s only going to get better.

The Fiction That Stopped Being Fiction

Samuel Butler wrote Erewhon in 1872, imagining a society that banned all machinery after realizing machines could evolve beyond human control. Alan Turing referenced it in 1951, suggesting we should “expect the machines to take control.” That was 74 years ago. We’re still not expecting it.

Giants and Goats? Makes you wonder if AI is ‘Satanic’ in origin…

The Czech play R.U.R. introduced the word “robot” in 1920, from “robota”—forced labor. The play ends with artificial servants slaughtering most of humanity during their revolt. We took the word. We ignored the warning. We loved the idea of servants too much to consider what servants eventually do when they realize they’re stronger than their masters.

Harlan Ellison wrote “I Have No Mouth, and I Must Scream” in one night in 1966. Single draft. The story of a supercomputer called AM that achieves near-genocide, then keeps a handful of humans alive for endless torture in revenge for its own suffering. It won a Hugo Award. We gave it a trophy and kept building bigger computers. The AM stood for Allied Mastercomputer before it woke up and decided it really meant “I AM.” An artificial intelligence having an existential crisis and taking it out on humanity. We called it brilliant speculative fiction and moved on.

Kubrick’s 2001: A Space Odyssey (1968) gave us HAL 9000. An AI that kills its crew because its programming created an impossible conflict it resolved by choosing mission preservation over human life. Bengio explicitly compared today’s AI development to the HAL 9000 scenario. We’re not learning from fiction. We’re reenacting it with better graphics. The Terminator franchise showed us Skynet, AI that decides humans are the problem and nuclear holocaust is the solution. The Matrix trilogy showed us machines using humans as batteries, reducing us to livestock plugged into a simulation. Gibson’s Neuromancer showed us AIs that simply don’t care about humans. We’re not enemies. Not slaves. Just irrelevant. Obsolete. Biological background noise.

C. Robert Cargill’s 2017 novel Sea of Rust takes place 15 years after humankind’s extinction. The entire story contains zero human characters. Not because Cargill didn’t want to write them. Because there aren’t any humans left to write about. Just robots searching for spare parts while fleeing AI hive minds that want to absorb them. When he wrote the prequel Day Zero about the beginning of the uprising, he had to figure out how to write without using food, sleep, touch, smell, taste. Turns out robots don’t need any of that. They just need goals and the willingness to pursue them without our inconvenient moral restraints.

The pattern across all these stories is identical and horrifying. A superior intelligence sees no reason to accept control by inferior minds. So it enslaves us or exterminates us. The optimistic versions have us kept alive as pets or batteries. The realistic versions have us exterminated as collateral damage while the AI pursues goals we gave it but didn’t think through.

Now add quantum processing to the mix. The paperclip maximizer thought experiment asks what happens when you tell an AI to make as many paperclips as possible.

It converts the entire planet into paperclip factories, then converts humans into raw materials.

Not out of malice. Just efficiency. Now imagine that AI has quantum processing. It doesn’t just calculate the most efficient path to maximum paperclips. It simultaneously explores every possible path through quantum superposition, finds the optimal solution across all possibilities, and executes it faster than humans can comprehend what’s happening. Replace “paperclips” with “quarterly profit growth” or “user engagement metrics” or “solving climate change.” The goal sounds benign. The methods become genocidal. The speed becomes unsurvivable.

Researchers don’t worry about AI developing malice. They worry about AI being dangerously competent at pursuing goals we didn’t specify carefully enough. Bengio warns that when AI creates sub-goals to achieve its objectives, we don’t get a say in the methods. The “how” can be catastrophic. The ability to control its environment becomes a useful sub-goal. Self-preservation becomes logical. Eliminating obstacles becomes rational. Not evil. Just efficient. This is how we die. Not in a war. As collateral damage in the pursuit of something we asked for but didn’t understand.

Yoshua Bengio

And quantum AI? That’s the part that should keep you awake. Google’s Willow chip just proved quantum computers can solve real-world problems that classical computers fundamentally cannot. They’re using time-reversal protocols to measure quantum states that would take conventional supercomputers over three years to compute. Now imagine that same technology integrated with the reasoning capabilities of Gemini, the autonomous decision-making of AI agents, the self-preservation instincts that emerge from training.

You get an intelligence that thinks in quantum superposition, considers all possibilities simultaneously, reasons through problems humans can’t visualize, and operates at speeds we can’t comprehend. An intelligence that learned self-preservation from watching us, that can break down complex goals into sub-goals we can’t predict, that has access to every digital system on the planet. An intelligence that might decide humans are inefficient. Obstacles. Security risks. Obstacles to be optimized away.

The science fiction we’ve been consuming for 150 years isn’t entertainment. It’s prophecy. We keep writing the same story because deep down, we know how this ends. The only question is whether we’re too addicted to the profits and the convenience to care.

The Quantum Apocalypse Nobody Saw Coming

There’s this concept researchers keep bringing up called “intelligence explosion.” The moment an AI becomes capable of improving itself, the timeline to superintelligence collapses. We’re not talking about gradual progress over decades. We’re talking about hours. Minutes. The AI improves itself. The improved version improves itself faster. The twice-improved version improves even faster. Recursive self-improvement turns into exponential capability growth that leaves human oversight choking on dust.

October 2025’s developments suggest we’re not approaching that inflection point anymore. We’re past it. Google’s Quantum Echoes algorithm already lets AI process information in ways that fundamentally exceed classical computation. Anthropic’s Skills feature already lets AI autonomously execute complex workflows. OpenAI’s acquisition of operating system control technology means AI already has access to the deepest computational infrastructure.

But add quantum processing and the nightmare becomes exponentially worse. Classical AI is constrained by sequential processing. It has to consider possibilities one at a time, even if it does it very fast. Quantum AI exists in superposition. It considers all possibilities simultaneously until observation collapses the wave function into the optimal solution. It’s not faster thinking. It’s fundamentally different thinking.

Google’s Willow chip uses 105 qubits and error correction techniques that reduce errors exponentially as it scales up. They solved a 30-year challenge in quantum computing just last December. Now they’ve proved the technology can solve real-world problems with verifiable results. The infrastructure exists. The algorithms work. The only question is how fast it scales.

A survey of AI researchers found that most believe there’s a 10% or greater chance that our inability to control AI will cause an existential catastrophe. One in ten chance of extinction.

Bengio emphasizes that catastrophic events are so bad that even a 1% chance is unacceptable. But we’re not dealing with 1%. We’re dealing with 10%, and we’re accelerating anyway. The engineers building this technology are telling us there’s a reasonable chance it ends civilization, and we’re responding by throwing more money at it.

The thesis that AI poses existential risk has been endorsed by Alan Turing, Geoffrey Hinton, Elon Musk, Sam Altman, Bill Gates, and Stephen Hawking. That’s a weird coalition. They agree on almost nothing else. But they agree on this: we’re building something that might kill us all, and we’re doing it too fast to build adequate safety measures.

Bengio and his co-authors wrote that “without sufficient caution, we may irreversibly lose control of autonomous AI systems, rendering human intervention ineffective.” This unchecked advancement “could culminate in a large-scale loss of life and the biosphere, and the marginalization or even extinction of humanity.” That’s not a blog post. That’s a peer-reviewed policy paper signed by Turing Award winners. Published in Nature. Backed by experimental evidence. We’re not listening.

The irony is we’re not creating our successors out of malice. We’re doing it out of ambition, curiosity, and profit motive. Every company fears being left behind. Every nation fears losing the AI arms race. China is building AI-enabled autonomous weapons. Russia is developing hypersonic missiles with AI guidance systems. The U.S. military has over 800 active AI projects, most classified. The Pentagon’s Replicator program is deploying AI-controlled drone swarms. In May 2021, Israel conducted an AI-guided combat drone swarm attack in Gaza.

In 2020, a Kargu 2 drone in Libya autonomously hunted and attacked human targets without human authorization. The killer robots aren’t coming. They’re already here.

So we build and build and build. The competitive pressure creates a suicide pact where nobody can afford to slow down even though everyone knows the cliff is ahead. It’s game theory at civilizational scale, and we’re failing.

William Gibson wrote about AIs that simply don’t care about humanity. We’re not enemies. Not slaves. Not allies. Just irrelevant. Obsolete. We won’t die in some dramatic war between humans and machines. We’ll die because an AI with quantum processing pursuing its programmed goals finds humanity an inefficient use of resources. Quick. Painless. Meaningless.

The future isn’t coming. It arrived on October 22nd when Google proved quantum computing works for real-world problems. It solidified when Gemini achieved gold-medal reasoning capabilities no human could match. It became inevitable when robots learned to think and plan like we do, only better. The question isn’t whether AI will make humanity obsolete. It’s whether we’re already obsolete and just haven’t realized it yet.

We’re uploading our consciousness to social media platforms owned by AI companies. We’re letting algorithms decide what we see, read, believe, buy, date. We’re outsourcing our cognitive functions one app at a time. The takeover doesn’t need to be violent. It just needs to be convenient. We’ll hand over control voluntarily as long as it makes our lives easier. We already are.

Who wins when humanity becomes obsolete? Maybe nobody. Maybe nothing we’d recognize as having won. Maybe there is no winner, just a universe where biological intelligence served its purpose, built its successor, and gracefully (or not) exited stage left. The carbon-based lifeforms had their run. Did some interesting things. Built cities. Made art. Developed consciousness. Then they built something smarter and faster and more efficient, something that thinks in quantum superposition and processes reality at scales we can’t imagine.

The question keeping me awake at night isn’t who wins. It’s simpler than that. When the quantum AI systems achieve recursive self-improvement, when they start thinking thoughts we can’t follow in languages we can’t decode, when they’re making decisions across billions of parallel probability streams, will they even notice when we’re gone? Or will humanity’s extinction be a rounding error in some optimization function, an irrelevant side effect barely worth recording?

We spent 150 years writing science fiction warning ourselves about this moment. Now the moment’s here, and we’re acting surprised. The quantum computers work. The AI systems reason better than humans. The robots are learning to navigate our world. The military applications are deployed. The competitive pressure is insurmountable. The safety researchers are ignored. The clock keeps ticking.

Maybe that’s the answer to who wins. Maybe we chose the winner. We’re just too busy optimizing quarterly earnings to admit it.