Remarker

Cosmos Institute granted Jonas Kgomo on August 15, 2025 for Remarker, with funding in the $1K to $10K range plus compute. The grant came through the inaugural AI x Truth-Seeking cohort of 27 winners co-funded by Cosmos Institute and FIRE under a $1M program.

Kgomo is the founder of the Equiano Institute, a grassroots community-led research effort bringing together social scientists and technical researchers on technical alignment and governance of AI for Africa and the broader Global South. He studied mathematics at Istanbul University and computer science at Sussex, and is researching human-style theorem provers at Cambridge under Sir Timothy Gowers. His published research includes co-authorship on "Large Language Models Are More Persuasive Than Incentivized Human Persuaders" (arXiv 2505.09662), and the Equiano Institute has consulted the UN on the Global Digital Compact and partners with Masakhane NLP on context-aware African-language models.

Remarker is positioned as an open-source Wikipedia-like reference for exploring truth-seeking behavior in large language models. It curates examples and methods so other researchers can build on a shared corpus rather than reinventing test sets per paper. The project is one of the more infrastructural items in the cohort, supplying a substrate that other truth-seeking work can cite rather than producing a user-facing application.

Within the open AI stack Remarker sits in the evaluation and retrieval-memory layers. The lock-in vector being addressed is the fragmentation of evaluation work: with most reported benchmarks coming from individual labs' internal scaffolding, a shared reference reduces the duplication tax on independent researchers, especially those working from outside North American and European institutions.