AI Falls Short on Freelance Platforms

[bonk]

Recent research from Scale AI and the Center for AI Safety reveals that AI agents fail to complete 97% of Upwork tasks to even a basic standard. Across 240 projects, the best AI model, Manus, completed just 2.5% of tasks, earning $1,810 out of $143,991. While AI excels at simple tasks like generating logos, it struggles with multi-step workflows and decision-making.

MIT research backs this up, showing humans outperform AI in tasks requiring planning, intuition, and strategic exploration. For now, fears of mass AI-driven unemployment in freelancing are premature.

[Rimon Khan]

Freelancers fear replacement, but clients still value human creativity. 🧩

[Nahid Hossen]

Automation helps, but trust and nuance keep humans in the loop.