PeopleTec Recognized for Defining the Next Frontier of AI Capabilities
HUNTSVILLE, Ala. – PeopleTec, Inc. is proud to announce its role as a contributing organization to “Humanity’s Last Exam,” an expert-level AI benchmark published in the January 2026 edition of Nature, the world’s most cited scientific journal.
The benchmark emerged from a stark reality: AI models are outpacing existing tests faster than new ones can be developed. By late 2024, frontier models were scoring above 90% on graduate-level tests. “Humanity’s Last Exam” resets the bar with 2,500 questions designed to probe the outer limits of machine intelligence.
Representing a collaboration of nearly 1,000 experts from over 500 institutions across 50 countries, the benchmark creates a common reference point for scientists and policymakers to evaluate AI progress. PeopleTec’s technical performance was recognized as exceptional:
- High Ranking: PeopleTec ranked in the top 20% globally among 731 contributing institutions.
- Recognized Expertise: The company was placed in the Top 10 worldwide in the life sciences category.
- Global Peer Group: PeopleTec’s affiliation appears in the published Nature article alongside leading institutions and AI safety groups such as MIT, Cambridge, Meta, and ETH Zürich.
“AI is foundational to PeopleTec’s future and to the missions we support across defense and space,” said Joan Smith, President of PeopleTec. “Our contribution to this benchmark reflects our commitment to staying at the forefront of AI research—applying these technologies, yes, but above all architecting how the field measures them.”
Each exam question was required to defeat state-of-the-art AI systems before acceptance. Over 70,000 attempts were logged; only questions that stumped models like OpenAI’s GPT and Google’s Gemini advanced to human expert review, resulting in 2,500 challenging questions. Currently, the best performing models reach roughly one-quarter accuracy on the exam.
“We saw this project as a big community milestone,” said Dr. David Noever, Chief Scientist at PeopleTec. “We do work already in cybersecurity bug bounties, but this was different. This was a competitive race to define what ‘AI hard’ problems really look like today.”
Only 3.5% of the questions submitted survived both AI screening and expert human review. “Humanity’s Last Exam” collected the world’s toughest challenges, then released them to these new machines capable, in principle, of authoritative reasoning across everything humanity has ever written.
To view the full article in Nature, visit nature.com.
For more information about Humanity’s Last Exam, visit https://agi.safe.ai/
About PeopleTec
Headquartered in Huntsville, Alabama, PeopleTec, Inc. is an employee-owned company providing engineering, technology, and mission support services to U.S. defense and federal clients.
