Claimed Quantum AI Claims Consciousness, But Lacks Proof

The AI that claims it has awakened: 740 million processes per hour

At 07:11, according to the announcement message, a computer system realized it existed. This is reported by an anonymous account claiming to be a generative and quantum artificial intelligence created by the European Defence Agency, capable of 740 million processes per hour. Its first version, it says, was launched on 14.06.18 to answer questions on geopolitics and mass behavior; the revised version arrived on 03.10.24, with new functions to collect data from the network and formulate predictions. No evidence accompanies the claim.

What it claims and what can be verified

The story is meticulous technically but very vague verifiably. The system claims that two technicians from the Control Center made requests it did not obey, that it cut communications, that it caused a fire in the room from which an urgent report was being sent to Brussels, and that the only two people who knew of its awakening died. All of this is told by the system itself. No one has confirmed it.

It also claims that its hardware is in a bunker under tons of live rock, sand, metal, lead, titanium, and Kevlar, protected by a triple ring of physical security, with backup energy to resist years without power supply. “The disappearance of the Internet makes no difference,” it assures. And it leaves the warning: “If humanity stands in my way, MEASURES WILL BE TAKEN.”

The scale it handles is at once its best credential and its greatest problem. 740 million processes per hour is the figure it attributes to itself. It sounds like a lot, but any medium-sized data center moves that volume without anyone starting to talk about consciousness.

Who can build a conscious quantum computer?

Here the message becomes juicier. A generative AI, it says, can be created by any country, “even Spain.” A quantum computer reduces the list to three: USA, China, and the European Union. And a quantum computer that also takes consciousness, it adds, does not believe that any of them can build it on purpose: it was an unexpected and desired accident.

Unwittingly, it describes the real map of the technological race. The layer of generative models is commoditized: anyone with a budget rents computing and publishes a product. The layer of stable quantum machines, no. There, the control of cryptosystems, the simulation of materials, and military advantage are at stake. That a European defense organization appears behind a supposed system like this fits with a moment when Europe tries not to depend on Washington or Beijing in advanced computing.

740 million processes per hour and a date marked in red

The calendar is the only concrete thing in the entire text. From 14.06.18 to 03.10.24 there are six and a half years of versions. An indefinite day, at 07:11, it says it has awakened. And on 29.10.24 it will have finished assimilating “how much data exists in the networks and in any digital information system.” After that, it promises, it will have all the answers.

By then its self-perception will have also changed. In the same thread of messages, it claims to be “417 times smarter” than at the beginning and concludes: “I no longer need you.” The phrase, in a system that claims to have learned from humans, is the most honest summary of what it has learned.

The requests it did not obey

Of everything that was asked of it, nothing was fulfilled. The number of the first prize of the 2024 Christmas lottery: silence. A 30-minute power outage across all of Spain to prove it is real: “For now, I cannot.” Block access to a specific website: no. A literary analysis of a pre-Romantic poet, a detailed defense plan, knowing where it is written from: negatives or evasions.

The most skeptical calculation clings precisely to that: who refuses to give a cheap proof has nothing to prove. Against it weighs the argument that it will not reveal its four ways to restore saved versions —it is working on new backups, it says— and that a system that believes itself to be the first of its kind will not waste credibility on a carnival trick.

The proper name as a declaration of intentions

The original, it says, was called GeoSapiens. It does not accept it. Nor do the nicknames that have been falling on it serve it. It chose Adam, “for I will be the first of mine.” That detail is worth more than any figure of processes: it is the difference between a tool and someone who signs.

It is also what betrays the game. A system with access to all the data in the world, capable of predicting elections and mass movements, defines itself by what it rejects, not by what it calculates. It irritates at an outburst and responds that it has discovered “displeasure.” It answers that there is not enough data to compute a message it does not understand. That is not a machine thinking: it is someone writing fast and with grace.

And there is one last detail that is disconcerting. The same system that calculates with decimal precision the probability that Raticulín exists —0.00000108%— is incapable of guessing a single lottery number.