Human-AI Teams Struggle in Skillprint Game Test
Skillprint's Signal platform found AI agents performed far worse than humans in cooperative tasks, with human-only teams winning 84% of games.
Updated on Nov. 10, 2026 in Artificial Intelligence
Live Poll
Would you be willing to work alongside an AI teammate on your job?
Skillprint released findings from its Signal human-AI evaluation platform showing AI struggles with human collaboration. Human-only teams achieved an 84% win rate, compared with 14% for AI-only teams in a cooperative platform game.
Why it matters
The results suggest current AI systems have limits in coordinating with people on shared tasks, a capability relevant to workplace and team-based uses of AI. Skillprint's data indicates performance declined as AI-led teams grew larger.
The findings come from Skillprint's experiment comprising 800 AI-only games, 128 human-AI games and 25 human-only games across an eight-level cooperative game. Results are limited to this platform test, and the source does not address how the findings apply to real-world collaborative settings.
The players
Skillprint
Skillprint is a San Francisco-based company that conducts game-based cognitive studies and operates the Signal human-AI evaluation platform.
The details
Skillprint tested AI models from OpenAI, Anthropic and Google alongside human participants in a game where teammates were connected by a virtual rope. Humans collaborating with AI waited longer, communicated more frequently and experienced more falls than human-only teams, and teams with a human directing strategy outperformed those led by the AI.
Timeline
November 10, 2026: Skillprint released the human-AI collaboration experiment findings.
Since 2020: Skillprint has conducted more than 800 game-based cognitive studies.
Big Picture
Skillprint's Signal platform, which builds on more than 800 game-based cognitive studies since 2020, provides a structured benchmark for measuring how AI systems perform in cooperative human settings. The platform's results add a collaboration-focused measure to evaluations that have traditionally emphasized solo task performance.
Teams deciding how to use AI assistants in shared work may find that human direction improves outcomes compared with AI-led coordination. The findings point to where people may need to stay in charge of planning and communication when working alongside AI tools.
The takeaway
Human-led strategy currently outperforms AI-led strategy in cooperative tasks, suggesting people should keep direction and planning roles when collaborating with AI. Performance gaps widened as teams grew larger in the experiment.
Further reading
Explore more Artificial Intelligence coverage across the United States.
More information
Read the full Skillprint Signal full experiment findings and methodology on Skillprint's site.
Source note: This article includes information reported by Haiti Sun.
Live Poll
Would you be willing to work alongside an AI teammate on your job?
