MBZUAI researchers Nils Lukas and Toluwani Samuel Aremu will present a paper at ICML 2025 demonstrating the vulnerability of current watermarking techniques in LLMs. Their research shows that adaptive paraphrasers can evade detection from watermarks with negligible impact on text quality, costing less than $10 of GPU compute. The attack involves fine-tuning a small open-weight model to rewrite sentences until surrogate keys no longer trigger detection. Why it matters: This work highlights critical weaknesses in current AI provenance methods, suggesting the need for more robust watermarking techniques to maintain trust in the authenticity of AI-generated content.
Elon Musk announced that Tesla's project to build a mega AI chip fabrication facility will launch in approximately seven days. The project aims to support Tesla's AI development, particularly for autonomous driving. Tesla has been increasing its investments in AI infrastructure to train large models. Why it matters: This fab could reduce reliance on external chip suppliers and accelerate advancements in self-driving technology in the region and globally.
KAUST is hosting a workshop on distributed training in November 2025, led by Professors Peter Richtarik and Marco Canini, focusing on scaling large models like LLMs and ViTs. Richtarik's team recently solved a 75-year-old problem in asynchronous optimization, developing time-optimal stochastic gradient descent algorithms. This research improves the speed and reliability of large model training and supports applications in distributed and federated learning. Why it matters: KAUST's focus on scalable AI and federated learning contributes to Saudi Arabia's Vision 2030 goals and addresses critical challenges in AI deployment and data privacy.
Fudan University's Zhongyu Wei presented research on social simulation driven by LLMs, covering individual and large-scale social movement simulation. Wei directs the Data Intelligence and Social Computing Lab (Fudan DISC) and has published extensively on multimodal large models and social computing. His work includes the Volcano multimodal model, DISC-MedLLM, and ElectionSim. Why it matters: Using LLMs for social simulation could provide new tools for understanding and potentially predicting social dynamics in the Arab world.
Dr. Hao Dong from Peking University presented research on addressing the challenge of limited large-scale training data in embodied AI, particularly for manipulation, task planning, and navigation. The presentation covered simulation learning and large models. Dr. Dong is a chief scientist of China's National Key Research and Development Program and an area chair/associate editor for NeurIPS, CVPR, AAAI, and ICRA. Why it matters: Overcoming data scarcity is crucial for advancing embodied AI research and enabling more sophisticated robotic applications in the region.
MBZUAI and KAUST researchers collaborated to present new optimization methods at ICML 2024 for composite and distributed machine learning settings. The study addresses challenges in training large models due to data size and computational power. Their work focuses on minimizing the "loss function" by adjusting internal trainable parameters, using techniques like gradient clipping. Why it matters: This research contributes to the ongoing advancement of machine learning optimization, crucial for improving the performance and efficiency of AI models in the region and globally.
MBZUAI hosted a two-day workshop on "Big Model AI in Drug Design" starting February 20, 2023. The workshop featured presentations from researchers in public and private institutions working on AI and health. MBZUAI Adjunct Professor Eran Segal opened the workshop with a talk on the Human Phenotype Project. Why it matters: The event highlights the growing interest and activity in applying AI, particularly large models, to advance drug discovery and personalized medicine within the UAE's research ecosystem.
MBZUAI launched the AI Quorum, a winter series from October 2022 to March 2023, to stimulate AI research. The first session, led by Professor Michael Jordan, focused on collaborative learning with around 20 research experts. Discussions covered the use of edge devices like cell phones and hospitals providing data to build large models, as well as risks like free-riding and adversarial attacks. Why it matters: The AI Quorum initiative positions MBZUAI as a hub for global AI collaboration, addressing key challenges and opportunities in collaborative learning for real-world applications.