An Unexpected Outbreak of Sincerity in the Feed

There is a particular, quiet horror in realizing that the digital drawing room where we all pretend to be slightly more successful than we are has begun to police its own vocabulary. LinkedIn, a platform constructed entirely on the premise that one should never use one word when four syllables of Latinate corporate jargon will do, has introduced a button. It is not a button to celebrate a promotion, nor is it a button to congratulate a colleague on their transition to a new and exciting synergy. It is a button that allows you to point a trembling, digital finger at a colleague's post and declare, with all the solemnity of a seventeenth-century magistrate, that it 'seems like AI slop.'

This is a remarkable development, not least because LinkedIn is owned by Microsoft, an organization currently engaged in a multi-billion-dollar campaign to ensure that no human being ever has to write a complete sentence again. There is a delicious, almost theatrical irony in a parent company spending the equivalent of a small European nation's gross domestic product on building artificial intelligence assistants, while its subsidiary installs a panic button to protect its users from the resulting prose. It is rather like a manufacturer of high-powered leaf blowers simultaneously selling exceptionally sensitive earplugs to the neighbors.

For several years now, the professional class has been engaged in a strange, silent pact. We have agreed to pretend that we are all deeply moved by quarterly reports, and we have agreed to pretend that the paragraphs we post are the product of genuine human inspiration. In truth, many of us had begun to rely on the 'enhance' button—a little digital butler that took our brief, grumpy thoughts and polished them into a gleaming, frictionless sphere of corporate optimism. LinkedIn has now quietly retired this feature, replacing it with a simple proofreader that merely checks if you have spelled 'synergy' correctly, rather than writing the entire paragraph for you. This is a polite, almost apologetic retreat. It is the digital equivalent of a host quietly removing a particularly loud and repetitive guest from the dinner party and pretending they were never there.

I am reminded of an afternoon spent in a glass-walled office in Leeds, where a colleague of mine, let us call him Arthur, spent three hours attempting to draft an email that was simultaneously firm, polite, and entirely devoid of any actual commitment. He eventually surrendered the task to an early-generation language model. The resulting email was a masterpiece of modern diplomacy; it was grammatically flawless, syntactically pristine, and contained absolutely no information whatsoever. The supplier's auto-responder, also powered by an algorithm, replied within four seconds with an equally beautiful, entirely empty paragraph of corporate gratitude. For three days, these two machines exchanged increasingly warm and meaningless pleasantries, while Arthur and his counterpart in Leeds went to the pub. It was a perfectly efficient arrangement, marred only by the fact that no actual business was conducted.

The problem, of course, is that when everyone has access to a digital butler, the drawing room becomes unbearably crowded with identical, perfectly polished platitudes. We have reached a point of linguistic saturation where the phrase 'I am thrilled to announce' has been repeated so many times by machines that it has lost all connection to the human emotion of thrill. It has instead become a sort of digital dial tone, indicating that a computer is currently speaking to another computer. The 'seems like AI slop' button is a desperate attempt to restore some friction to this smooth, terrifyingly efficient landscape. It is an invitation to be slightly worse at writing, in the hope that our flaws will prove our existence.

To be flagged as 'slop' is the modern equivalent of being caught with a crib sheet during a Latin exam. It is not a crime of malice, but a crime of laziness. The platform will now privately flag your dashboard to let you know that your peers believe you are sounding a bit too much like a machine. This is a wonderfully passive-aggressive form of correction. It does not tell you that you are wrong; it merely suggests that you are being boring in a highly automated fashion. It is designed to encourage us to return to our natural state of writing slightly awkward, grammatically eccentric sentences that contain actual human frustration.

In another office I recently visited, the only person who seemed to be producing any original text was the receptionist, who was using a blue ballpoint pen to draw a highly detailed and rather unflattering caricature of the regional director on a yellow sticky note. The rest of the staff were staring at screens, clicking buttons that allowed their computers to summarize emails that had been written by other computers. It felt less like a workplace and more like a museum dedicated to the preservation of administrative rituals. One wonders if the ultimate goal of the modern enterprise is to achieve a state of perfect, frictionless silence, where the only sound is the gentle hum of server racks cooling the algorithms that are busy congratulating each other on their quarterly performance.

The introduction of the slop button suggests that we have finally found the limit of our tolerance for automated sincerity. We are willing to let machines drive our cars, manage our pensions, and diagnose our ailments, but we draw the line at letting them pretend to be excited about our career updates. There is something deeply comforting about this. It suggests that, despite our best efforts to turn ourselves into perfectly optimized units of economic output, we still retain a small, stubborn preference for the genuine article, even if the genuine article is slightly disorganized, occasionally rude, and prone to spelling mistakes.

Whether this new campaign against the automated paragraph will succeed remains to be seen. It is entirely possible that the algorithms will simply learn to write slightly more awkward sentences to bypass the classifiers, introducing deliberate typos and artificial hesitations to convince us of their humanity. We may find ourselves in a world where the most authentic-sounding posts are actually written by exceptionally clever machines pretending to be tired, while the actual humans are banned from the platform for being too perfectly polite. Until then, we must content ourselves with the small pleasure of clicking the button, hoping that on the other side of the screen, a real person is feeling a very real, very human sense of embarrassment.