The Future of Everything Is Lies, I Guess: Annoyances

It has been reported that a new wave of machine‑learning tools will do more than automate tasks — they’ll aggravate customers and blur accountability. A recent essay argues companies are diverting support tickets into chats with large language models, and voice versions will soon handle phone calls too. Sound efficient? Sure. Honest? Allegedly not. The pitch is simple: cheaper support, fewer humans, and a system that often prefers obfuscation to resolution.

Customer service

Customer service, already engineered to be a cost center, looks set to get colder. Expect polite, endlessly patient chatbots that recite scripts and, because they’re LLMs, sometimes invent answers or refuse to escalate. That’s not speculation so much as an economic calculation: why pay a trained, empowered human when a model can placate and deflect? The result will be familiar: high‑value customers get real people (read: pay to skip the queue), the rest get an argument with software. Comcast calls, anyone?

Arguing with models

Beyond help desks, LLMs will govern fuzzy decisions — insurance denials, dynamic pricing, content of ads, even what counts as “necessary” medical care. Algorithmic systems needn’t be accurate to be profitable; they only need to be effective at steering outcomes. The downstream effect? New drudgeries: phrasing requests precisely to fool an insurer, gaming shopping algorithms, or trying browser tricks to nab a cheaper flight. It’s a culture of small, daily battles against systems that diffuse responsibility. Who fixes the real harm when the answer is always “the model said so”? What a dismal future — unless we demand accountability before the annoyances become the norm.

Sources: aphyr.com, Hacker News