American political scientist Francis Fukuyama is best known for declaring the “end of history” after the Cold War. History, of course, had other plans.  

Now, Fukuyama has discovered a deeper wonder: the consolation of artificial friendship. In a recent column, he recounts how ChatGPT helped him migrate his sprawling personal database, taught him proper coding habits, and cheered him on with a steady stream of affirmations. By the end, he was less a user than a grateful pupil. 

It is a future that Alan Turing, mathematician, Enigma codebreaker and WWII hero, would hardly recognise. And perhaps, if he had, he might have quietly unplugged. 

Turing proposed the “Turing Test” as a crowning achievement for machines. 

Success would come when they could imitate the rich, muddled quirks of human conversation so well that no one could tell the difference. 

In 2025, we have achieved something nobler still: the complete reversal of the challenge. Machines no longer strive to reach a human level of proficiency. We, it seems, are struggling to meet theirs. 

Chatbots reply with flawless grammar, infinite patience, and a generous refusal to indulge in spurious conspiracy theories. Humans, meanwhile, bicker, forget, misquote, misplace commas, and explain with rising passion things they plainly do not understand. 

Identifying the machine is easy: it’s the one that still makes sense (the odd ‘hallucination’ notwithstanding). 

Developers now face a new challenge: downgrading their chatbots to make them more human. New “Authenticity Modules” could reintroduce random typos, the occasional logical fallacy, stubborn wrong answer, mistimed emotional outburst, or wild accusation in the middle of an otherwise sensible conversation.  

A “premium version” could allow the chatbot to forget basic geography, confuse historical wars, and start furious arguments about topics it barely understands.  

Thus, the Turing Test survives – not as a measure of artificial achievement but as a delicate exercise in artificial degradation. Passing for human now requires careful calibration: just enough error, just enough confusion to be believable. 

We feared the machines would get too good. We never imagined how easy we would make it for them. 

Turing hoped to establish a criterion for machine intelligence. Instead, it established a monument to human decline – a test that machines must now deliberately fail to keep our pride intact. 

The future is here. It is just a little bit dumber than expected.