You already have an account? Log In

Task Paralysis and AI: Why More Options Can Mean Less Action

Posted on 10.05.2026

There's a strange irony hiding inside our shiny new productivity stack. The same technology that can help a paralysed man move a robotic arm with his thoughts is, for many of the rest of us, quietly making it harder to start a simple email.

On one end of the spectrum, researchers are using AI to decode neural signals so people who cannot move their bodies can move machines. On the other end, knowledge workers with full use of their limbs are sitting at their desks, ChatGPT tab open, unable to decide which of seventeen suggested email drafts to send. Welcome to the age of task paralysis.

The new kind of stuck

Recruitment and tech-careers site Dice recently flagged what it called AI's hidden cost: career paralysis — a creeping sense among workers that with so many AI tools, prompts and possible workflows, it's getting harder to commit to a direction at all. Should you specialise? Generalise? Learn LangChain? Pivot into prompt engineering? Wait six months until the landscape shifts again?

That same dynamic plays out in miniature dozens of times a day. You ask an AI assistant for "five options" and get five plausible-sounding answers, none clearly better than the others. You ask it to help plan your week and it produces a 14-step schedule you will not follow. You ask it to write a message and end up rewriting the rewrite.

Psychologists have a name for this older than any chatbot: choice overload. The more options on offer, the more cognitive effort required to evaluate them, and the more likely we are to defer the decision entirely. AI didn't invent the problem. It just industrialised it.

Why AI amplifies the freeze

Three things make AI particularly good at triggering paralysis, even when it's genuinely useful.

1. It removes the friction that used to force decisions. When generating a draft took 20 minutes, you sent the draft. When it takes 20 seconds, you generate six and compare. The cost of producing options has collapsed; the cost of choosing between them has not.

2. It dissolves the edges of a task. "Write a project update" used to be a finite job. Now it can balloon into "write a project update, then ask the AI to critique it, then ask for three alternative tones, then research what your competitors do, then draft a follow-up plan." Every task has a trapdoor into a bigger task.

3. It produces fluent answers regardless of fit. Large language models are confident by design. Two contradictory suggestions can both sound reasonable, which means the human still has to do the hard work of judgement — but now wrapped in the false comfort that "the AI said so."

A useful contrast: AI that removes paralysis

It's worth pausing on the medical research, because it points to what AI does best. According to Scimex and a study covered by Nature, an AI-powered brain-computer interface (BCI) recently allowed a paralysed man to control a robotic arm using only his thoughts. As Science Media Centre España notes, the AI's job is to interpret messy neural signals and translate them into a single, clean instruction: pick up the cup, turn the wrist, release.

That's the template. The AI doesn't offer the user a menu of seventeen possible arm movements to choose from. It collapses noisy input into one decisive output. The human supplies intention; the machine handles execution.

Compare that to how most of us use generative AI at work. We supply the vague intention ("help me with this"), and the machine hands back a buffet of execution options for us to sort through. The roles are reversed — and so is the outcome. One frees a person to act. The other can quietly trap them in deliberation.

Five ways to use AI without overthinking

None of this means abandoning the tools. It means using them more like the BCI and less like a brainstorming partner with infinite stamina. A few practical habits help.

Decide the decision before you prompt. Write down what you actually need to walk away with: one email, one paragraph, one yes/no. If the output isn't that shape, ignore it. Treat the prompt as a request for an answer, not a discussion.
Ask for one option, not five. "Give me three versions" feels safer but reliably triggers comparison spirals. "Give me your single best version" forces the model — and you — to commit. You can always regenerate, but you'll usually accept the first.
Set a timer, not a quality bar. Decide in advance: "I'll spend 15 minutes with the AI on this." When the timer goes, send what you've got. Output quality from AI plateaus quickly; your willingness to ship doesn't have to.
Use AI to narrow, not to expand. The freeze usually comes from divergent prompts ("brainstorm ideas for..."). Try convergent ones instead: "Here are my three options. Which has the biggest weakness, and why?" You're using the model to eliminate, not to multiply.
Separate the thinking step from the doing step. If you're using AI to plan your week, close the tab before you execute. Don't keep re-consulting it mid-task. The Dice piece on career paralysis hints at the same idea on a longer timeframe: at some point you have to stop surveying the landscape and start walking across it.

The career-scale version

Zoom out and the same logic applies to the bigger questions. Australian workers — particularly in tech, marketing, design and admin-heavy roles — are being told daily that AI will transform their jobs. The sensible response, on paper, is to learn, adapt, experiment. The actual response, for many, is to freeze: too many courses, too many tools, too many possible futures, and a quiet suspicion that whatever you pick will be obsolete by Christmas.

The Dice analysis frames this as the hidden cost of the AI boom. It's not that people don't want to upskill. It's that the menu is too long and changes too fast to feel safe choosing from.

The way out is unglamorous: pick one tool, one workflow, one problem in your actual job, and use it badly for a month. Real competence is built from finished tasks, not researched options. The professionals who'll do well aren't the ones with the most comprehensive view of the AI landscape — they're the ones who shipped something with it this week.

The takeaway

AI is genuinely extraordinary. It can translate brain activity into movement for someone who has lost it. It can also, if we're not careful, translate a simple decision into an afternoon of comparison.

The fix isn't to use it less. It's to use it more like a clinician would: with a clear question, a tolerance for one good-enough answer, and the discipline to stop consulting and start acting. The tools are powerful. Whether they free you or freeze you is, mostly, a question of how you hold them.