UAE teams win top prizes in TII and Wazoku AI challenge

The Technology Innovation Institute and Wazoku named winners of the CrowdLabel Challenge, which drew 70 submissions from 29 countries to find scalable alternatives to human data labelling for large language models. Five of the winning contributors are based in the UAE, with the top prize going to a team behind a learning app that folds labelling into natural user interactions.

Why it matters: - The challenge targets a core bottleneck in AI: high-quality human data labelling for large language models. - Lower-cost, scalable labelling methods could reduce reliance on expensive expert annotators used in reinforcement learning with human feedback. - The results show open innovation can surface AI talent outside traditional tech hubs, including the UAE.

What happened: - The Technology Innovation Institute, part of Abu Dhabi’s Advanced Technology Research Council, announced winners of the CrowdLabel Challenge with Wazoku via the Innocentive platform. - The competition drew 70 submissions from people in 29 countries. - Five winning contributors were based in the UAE. - The Best Overall award and $30,000 went to UAE-based team Poornima Nair, Madihah Anaum and Megha Manoj for Leap Language App. - The Best Idea award and $5,000 went to Aditya Chatterjee, Adam Ibrahim and George Ibrahim for FlyQ. - Faidh Feisal of the UK won the Best Execution award and $15,000. - Vicky Kamau of Kenya and Amgad Samy of Egypt each received $3,000.

The details: - Leap Language App is designed to integrate data labelling into natural user interactions to generate training data at scale. - The Leap team said the project aims to turn AI into a tool that improves learning rather than distracts from it. - The Leap team said the project was their first major endeavor and that they wanted to inspire more young women to enter tech. - FlyQ was recognized for scalable, crowdsourced human labelling with high-quality outputs. - The FlyQ founders are aged 20 to 27 and have prior hackathon experience. - The FlyQ founders also built Socia, a UAE-based platform that connects innovators with opportunities. - Faidh Feisal’s winning idea turns digital advertising into data labelling opportunities. - Wazoku’s Innocentive platform connects organisations with more than 700,000 problem solvers across AI, engineering, science and other fields. - The competition focused on finding cost-effective alternatives to traditional RLHF methods. - TII framed the challenge as part of its push to advance AI research and development for next-generation intelligent systems. - The initiative also aimed to reduce costs, increase scalability and broaden participation in AI development worldwide.

Between the lines: - The strong showing from UAE-based winners suggests the region is becoming a more visible source of applied AI ideas, not just a market for imported tools. - The competition’s emphasis on crowdsourced and behavior-based labelling points to growing interest in data creation methods that can scale beyond elite expert review. - Wazoku’s open-innovation model appears to be a way for enterprises to tap global talent quickly and cheaply while widening the pool of potential AI contributors.

What's next: - TII and Wazoku are likely to keep using challenge-based sourcing to identify practical AI approaches that can be tested and scaled. - The winning ideas may now serve as prototypes or reference points for future work in AI training data generation. - The broader push to diversify who contributes to AI development is expected to continue as the sector looks for cheaper and more scalable model-training inputs.

The bottom line: - The CrowdLabel Challenge turned a niche AI infrastructure problem into a global competition, and UAE talent emerged as one of the clearest winners.