Task2k Policy on Using AI for Tasks

Task2k exists to give companies real, human insight into how people actually experience their apps and websites. That insight is only valuable if it's genuine. As AI tools have become easier to access and harder to detect, we've put together a clear policy on where they fit, and where they don't, in how testers use Task2k. This document explains what's allowed, what isn't, why the policy exists, and what happens if it isn't followed.

1. The Short Version

Do not use AI to generate your responses, opinions, or reactions during a study. Every answer, spoken or written, should come from your own head, in your own words, based on what you actually experienced while completing the task. AI tools can be useful in other parts of your life and work, but they have no place standing in for your genuine reaction as a tester. That's the core of the policy. Everything below explains the reasoning and the edge cases in more detail.

2. Why This Policy Exists

Companies come to Task2k because they need to know how a real person, with real habits, real expectations, and a real reaction, experiences their product. That's the entire value we provide. An AI tool doesn't use the product, doesn't feel confused when a button doesn't do what it expected, and doesn't have a lived history of using similar apps that shapes how it judges a design. It can only generate a plausible-sounding answer based on patterns in text it was trained on, not on lived experience.

If AI-generated responses started making their way into studies, two things would happen. First, the feedback itself would get worse: generic, smoothed-over, and disconnected from what actually happened during the task, which is exactly the opposite of what makes usability testing useful in the first place. Second, and just as importantly, it would break the trust that makes the whole platform work. Companies pay for and rely on Task2k because they trust that the people behind the screen recordings and written answers are real testers having real reactions. If that trust erodes, it affects every tester on the platform, not just the ones using AI, because companies would have no way to tell genuine responses apart from generated ones, and might start doubting all of them.

3. What Counts as Misuse

To be specific, here's what is not allowed during a study:

• Using ChatGPT, Claude, or any other AI chatbot to write your typed answers for you. Even if you paraphrase or lightly edit the output afterward, if the core content and reasoning came from an AI tool rather than your own thinking, that's a violation.
• Reading from an AI-generated script while narrating your screen recording. The narration is supposed to capture your live thought process. Scripting it in advance with AI defeats that purpose even if you say it out loud in your own voice.
• Asking an AI tool what a "good" answer would look like and then reproducing it. This includes asking an AI to guess what a researcher wants to hear. There is no correct answer we're looking for; there's only your honest reaction, so there's nothing for an AI to correctly guess.
• Using AI to complete the actual task on your behalf, for example, asking an AI assistant to find information on a website for you and then reporting back what it found as if you found it yourself. The task needs to be completed by you, not delegated.
• Using AI-generated voice tools or avatars to stand in for your own voice or face where a study requires you to be visibly or audibly present.

4. What Is Fine

This policy isn't about AI in general; it's specifically about AI standing in for your genuine reactions during a study. A few things that are perfectly fine:

• Using AI tools in your normal life, work, or other freelance projects outside of Task2k. This policy only applies to what you submit as part of a Task2k study.
• Using a spell-checker or basic grammar tool on a written answer after you've written it in your own words, as long as it isn't rewriting your reasoning or generating new content.
• Asking Task2k support (which may itself use AI-assisted tools on our end) general questions about how the platform works, how to use the recording software, or how to fix a technical issue. That's unrelated to the substance of your test responses.
• Using accessibility tools you rely on day to day, such as screen readers, dictation software, or text-to-speech, where those tools are simply helping you interact with your device rather than generating your opinions or reactions for you. If you're unsure whether a tool you rely on falls into this category, reach out to us and ask before starting a study.

5. How We Check

We don't expect testers to memorize a long list of technical detection methods, and we're not going to lay out exactly how our review process works, but it's worth knowing that submissions are reviewed, and AI-generated text and speech patterns tend to be noticeably different from natural, in-the-moment human reactions. Overly polished phrasing, unnaturally even pacing, generic observations that could apply to almost any website, and answers that don't match what's actually visible in the recording are all things that stand out during review. Companies reviewing studies notice this too, and flagged submissions get looked at closely.

6. What Happens If the Policy Isn't Followed

We'd rather testers understand this policy up front than find out about it after a problem, which is exactly why we're publishing it clearly. That said, if a submission is found to violate this policy:

• The submission may be rejected, meaning it won't be paid out and won't count toward your completed studies.
• Repeated violations can result in a warning on your tester account.
• Continued or serious violations can result in suspension or removal from the Task2k platform.

We take this seriously because the integrity of every tester's work depends on it. A platform where some testers can shortcut the process with AI while others put in genuine effort isn't fair to the honest majority, and it isn't good for the companies relying on all of us.

7. Why Your Real Reaction Is More Valuable Than a Polished One

It's worth saying this plainly: a slightly messy, natural, honest answer is worth far more to a company than a smooth, well-structured, AI-generated one. Real hesitation, real confusion, real moments of "wait, where did that go," these are the exact signals that reveal where a product design breaks down. Companies aren't hiring testers because they want good writing; they're hiring testers because they want the truth about how a real person experiences their product. Every time you narrate a genuine dead end or explain a reaction in your own words, even imperfectly, you're doing exactly what makes Task2k valuable. Don't feel pressure to sound polished. Sound like yourself.

8. If You're Ever Unsure

If you're ever unsure whether something crosses the line, a simple test is to ask yourself: did this answer come from what I actually thought and felt while doing the task, in my own words, or did it come from somewhere else? If it's the former, you're fine. If you're still not sure, reach out to Task2k support before submitting rather than after. We'd always rather answer a quick question up front than review a flagged submission later.

Quick Summary

• Do not use AI to write your typed responses or script your narration.
• Do not let AI complete tasks on your behalf.
• Do use AI freely in the rest of your life; this policy only covers what you submit to a Task2k study.
• Basic spell-check and accessibility tools you already rely on day to day are fine.
• Submissions are reviewed, and violations can lead to rejected work, warnings, or account suspension.
• Your honest, unpolished, first-person reaction is the most valuable thing you can give a company, and it's the entire reason Task2k exists.