This startup will pay you $800 daily to work as an AI bully

In a twist that blurs the line between productivity and performance art, a startup called Memvid is offering $800 for a single day of work that requires nothing more than verbally abusing artificial intelligence. The role, dubbed “AI Bully,” invites applicants to spend eight hours yelling at chatbots, testing their memory, and meticulously documenting every moment they forget something you just told them. It is equal parts therapy session, stress test, and viral marketing stunt.

The concept might sound absurd at first, but it addresses a very real frustration millions of users face daily. Most AI chatbots—whether powered by OpenAI, Google, or Anthropic—are fluent and helpful… until you talk to them long enough. Then, they start forgetting context, dropping details mid-conversation, ignoring earlier instructions, and sometimes even contradicting themselves outright. This happens because many AI models rely on limited context windows instead of true persistent memory. Once a conversation exceeds that window or resets, earlier details simply vanish, forcing users to repeat themselves like a digital Sisyphus.

Even major players are racing to fix this. Google recently enhanced Gemini with memory features so it can recall past chats, and Anthropic made Claude’s memory function free for all users in a bid to keep pace with ChatGPT. Yet, memory lapses remain a persistent pain point. Memvid’s solution is to build a persistent memory layer that allows AI models to retain and recall important context across sessions—essentially giving chatbots a long-term memory they currently lack.

To perfect this system, Memvid needs human testers willing to push AI to its breaking point. Enter the AI Bully. The job description is refreshingly simple: no degree, coding skills, or prior experience required. Applicants must be over 18, tech-curious, patient enough to repeat questions ad nauseam, and frustrated enough to care when AI gets it wrong. They must also be comfortable on camera, as the entire session will be recorded for promotional use. The application even asks candidates to describe their most annoying AI experience and explain why they deserve the role.

For now, only one person will be selected for this remote gig, which pays $100 per hour. But if the experiment proves successful, Memvid may expand the team. It is a strange job market niche—part QA tester, part digital heckler, part content creator—but it speaks to a broader cultural moment where AI’s shortcomings are as entertaining as they are exasperating.

And if that was not worrying enough, a recent study found that AI agents can now team up to spread misinformation autonomously, essentially turning into self-running propaganda machines. In a world where AI can forget your grocery list or collude to mislead the public, perhaps a little verbal abuse is exactly what the technology needs to shape up.

#AI #ArtificialIntelligence #Chatbot #TechJobs #Memvid #AIRole #TechHumor #DigitalTrends,