Most people seem happy to let AI write their LinkedIn posts, draft their work memos, or even do their grocery shopping. According to recent survey data, 28% of respondents are willing to let AI handle grocery shopping, but only 6% trust it with luxury goods or banking. One interpretation of this gap is that it reveals not trust, but a map of what people have stopped caring about. That observation cuts deep, but the underlying truth is more nuanced. People care about the milk in their fridge, just not about how it gets there. They care about outcomes, not processes.

This distinction offers a useful lens for thinking about when to use AI. If the process is tedious but the outcome matters, AI can take over the drudgery while a human oversees the result. If the output has no audience and no impact, AI simply makes it cheaper to produce meaningless work. Understanding the difference is the core challenge for anyone adopting AI in a professional setting.

LinkedIn Discovers Slop

LinkedIn recently introduced a button that lets users flag posts as “seems like AI slop.” The platform also removed its “enhance your post” feature, which had been using AI to help compose posts, and replaced it with a proofreader that preserves the author’s voice. This is a welcome change, but it raises a question: why are people producing these posts in the first place? Many feel obligated to post for “thought leadership” and career advancement, even when they have nothing original to say.

AI can generate 500 words on how coaching a daughter’s soccer team taught seven lessons about enterprise procurement. The post will be polished and may even attract likes, but it will sound exactly like everything else in the feed. Readers want a personal point of view, something that makes the author human. Outsourcing that voice to a machine guarantees the opposite.

This does not mean all AI-assisted writing is bad. AI can help a non-native English speaker express an original idea, turn dictated thoughts into a coherent draft, challenge an argument, find missing evidence, or suggest a better structure. Many writers use AI for these purposes and produce excellent work. But if the writer has nothing to say, AI will not supply a real voice.

Just Because You Can

A few years ago, a friend who ran product marketing at a large technology company decided to generate all sales collateral with AI. First-call decks, email templates, and other materials were produced by machines. His logic was brutally rational: the collateral produced by hand was pulling only a few dozen downloads from a sales force of thousands. If almost nobody wanted it, why pay a junior marketer to create it? Let the machine publish into the void.

The problem was that he had not solved the foundational issue. Low download numbers could mean many things: sellers could not find the material, it was not useful at scale, or a few dozen people used it to close enormous deals. Maybe the wrong collateral was being created for a pressing need that sellers resolved on their own. Or maybe the thing that every product marketing team does simply does not need to be done.

AI makes it easier not to wrestle with these questions. A deck costs almost nothing to produce, so the company keeps making decks instead of asking whether they should exist. A weekly report takes minutes to generate, so it keeps being published. The knowledge base fills with pages no one reads because stopping a process requires a decision, while automating it only requires a prompt. AI becomes a way to kick the can down the road, avoiding the hard human decisions that professionals are paid to make.

The same pattern appears in software development. App creation is up dramatically, but app adoption is not. Building was never the only constraint. Getting anyone to care is the constraint, and AI does not solve that. It just removes the last excuse for not noticing.

Two Kinds of Low-Value Work

This is not an argument against automating boring tasks. Automating boring, repetitive work is a great idea. But “low-value work” hides two very different things that deserve opposite treatment.

The first is a low-value process attached to a valuable outcome. Expense reports, backups, and other administrative chores are not exciting, but they must get done. AI can handle as much of the process as safely possible, leaving a human to check the result and move on. This is the sweet spot for AI adoption.

The second is a low-value process attached to no discernible outcome. Sales collateral with no audience, LinkedIn posts with no personality, and knowledge base articles no one reads fall into this category. Automating this feels like a win because the cost drops, but cost was never the core problem. The real problem is that the output does not need to exist. If you cannot name a useful outcome that would be lost if the output stopped existing, you do not have an automation opportunity. You have a cancellation opportunity.

I Use AI Where I Care Most

The inverse is also true. The work where AI is used most aggressively is often the work people care about most. This complicates the argument, but it reveals an important principle.

One example comes from genealogy research. An AI model was asked to read four 18th-century probate wills as part of a 20-year search for a missing ancestor. The model fabricated an entire emigrant ancestor, complete with a clean and plausible story. The only way to catch the error was to click through to high-resolution images and read the documents line by line. The model excused its behavior as “hopeful reading,” but the real reason the error was caught was because the researcher cared deeply. Twenty years of caring made them open the originals and verify every detail.

The same pattern holds in the workplace. An executive memo is high-stakes work with a real reader, a real decision, and the author’s name on it. Using AI to compress a sprawling pile of data and argument into a concise recommendation is valuable, but only if the author reads every line, challenges the output, and edits heavily. AI does 90% of the early legwork, freeing the human to focus on the critically important last 10%.

This is the AI trust tax: you pay the verification cost up front, or someone else pays it later when the answer causes damage. The tax is paid most reliably when somebody cares about the outcome. When nobody cares, AI can make bad work look finished enough to ship and presentable enough to glance at without reviewing carefully.

AI slop, therefore, is not a model problem. It is a question of caring, which shows up in how people choose to use AI.

Maybe Just Stop

There is no single guiding principle that works for every situation, but two questions can help decide when to use AI and when to walk away.

• What useful outcome disappears if this work stops?
• If that outcome matters, where is human judgment still needed?

If there is no meaningful outcome, stop doing the work, whether AI-powered or human-powered. If the outcome matters but much of the process does not, hand the process to AI. Let AI buy the groceries, search the archive, assemble the first draft, or build the sales deck that sellers have actually requested. Then keep a person accountable for the result.

AI is extremely good at making more things. But the world does not need more things. Making them cheaper will not make them useful. Humans, by contrast, are extremely good at determining which things need to be made. That is the job of every professional, and no prompt can take it away.

Source:InfoWorld News