Humanity's Sixth Sense Benchmark Reveals AI Visual Reasoning Shortcomings
- Published
- Source
- Substack.com
Elorian and Scale AI have launched a new benchmark called Humanity's Sixth Sense (HSS) to evaluate AI models' intuitive visual reasoning capabilities. The benchmark consists of 522 questions based on 288 images and 234 short video clips, covering various reasoning categories. While human participants achieved a high aggregate pass rate of 93.1, the best AI model, GPT 6 Astra, scored only 53.6, indicating significant gaps in AI performance compared to human reasoning.
The significance of HSS lies in its potential to challenge existing assumptions about AI's capabilities in visual reasoning. Many experts believed that advanced spatially aware models would be necessary for such tasks. However, the performance of current large language models like GPT and Claude suggests that they can handle these tasks better than previously thought, raising questions about the future development of AI systems.
Looking ahead, the AI race is extending into space, with the 2030s expected to witness a surge in the space economy. As AI technologies are integrated into spacecraft design and manufacturing processes, there are concerns about the implications of AI dominance in space. The potential for AI to create compounding advantages in space raises critical questions about power dynamics and access to space resources.
Original report
Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space
Substack.com · Jack ClarkThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
-
1
RobotToday Weekly October 5 – 9, 2026 RobotToday · 2 days ago
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
Related Suppliers
Exomys – Augmented Humanity GmbH
Austria
Vienna, Austria exoskeleton startup registered 2018 by Johann Bernhardt and Veronika Pauser; sells the passive Atlas and Herkules support systems.
IEEE
United States
World's largest technical professional organization for electrical and electronics engineers; develops key robotics and automation standards.
Robocurve
United States
San Francisco public benefit corporation founded 2026 that benchmarks frontier AI models on real robot arms and maintains the Inspect Robots framework.
OpenAI
United States
US AI research lab and product company founded 2015; develops GPT, DALL-E, and robotics/embodied AI foundation models targeting physical-world AGI.
Tera AI
United States
San Francisco AI startup providing software-only spatial reasoning and visual navigation for robots, enabling GPS-free positioning using onboard cameras.
Shield AI
United States
US defense AI company building Hivemind — an autonomous AI pilot for military aircraft and drones operating without GPS or comms links.
Light Robotics
China
Singapore- and Shenzhen-based embodied AI startup founded 2023 by ex-OpenAI lead Jiang Xu, developing companion robots for home services and healthcare.
United Space Structures
United States
US startup developing robotic construction systems for large-scale permanent structures in lunar lava tubes using AI, robotics, and automation.
Liquid AI
United States
MIT-spinout building device-native foundation models (LFMs) that run on-device across phones, cars, robots and edge hardware.
Xiamen Reconova Information Technology Co., Ltd. 厦门瑞为信息技术股份有限公司
China
Reconova is a Xiamen vision-AI company that listed on the Hong Kong Stock Exchange in 2026 and builds airport baggage-handling robots.
Anthropic
United States
AI safety and research company and public benefit corporation that builds the Claude family of large language models.
Mentee Robotics
Israel
Israeli AI-first humanoid robotics company building MenteeBot, a vertically integrated bipedal robot that learns from human demonstration.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
