Alibaba's DeepSWE Dominates Global Benchmarks, Cursor AI Expands in Pakistan, and ChatGPT Surprises in Space Simulation

DeepSWE's success in the SWEBench-Verified test underscores Alibaba's expanding presence in the global open-source community. Developed by Agentica and Together AI, DeepSWE was trained using Alibaba Cloud's Qwen3-32B model, which belongs to Alibaba's third-generation AI model family. The open-source structure of these models allows developers to access, modify, and improve their capabilities, facilitating innovation and resolution of potential issues. (Source)

Cursor Pakistan Community held its first meetup at Islamabad's National Incubation Center, organized by local ambassador Yahya Qureshi, to foster an AI-driven coding ecosystem in Pakistan. The event united academia, developers, and students, featuring Cursor's smart code completion and AI-powered refactoring tools. Juan Batista from Anysphere Inc. joined virtually to discuss Cursor's architecture and integration practices. With plans for more events in Karachi and Lahore, this meetup underscores Cursor's commitment to empowering Pakistani developers with AI-native coding tools, reflecting Anysphere Inc.'s broader strategy following Cursor's rapid growth since its January 2023 launch. (Source)

Researchers from MIT and the Universidad Politécnica de Madrid tested OpenAI's ChatGPT as an autonomous agent controlling a simulated spacecraft and found it performed unexpectedly well, landing second in a competition based on the Kerbal Space Program. This competition, called the Kerbal Space Program Differential Game Challenge, explored different autonomous agents' methodologies, including reinforcement learning. Despite its success, ChatGPT's occasional hallucinations highlight the current limitations and risks of using AI models in space missions. Notably, this research was conducted prior to the release of OpenAI's more advanced GPT-4 model, indicating potential for even better performance with newer AI iterations. (Source)

Aalto University's Intelligent Robotics group at the School of Electrical Engineering seeks postdoctoral researchers for projects in robot learning, manipulation, navigation, and AI for physical systems. The Finnish research group, part of the Finnish Center for Artificial Intelligence and ELLIS Institute Finland, provides cutting-edge robot platforms including Franka Emika Panda, Boston Dynamics Spot, and an autonomous car named Alex. Candidates should hold a PhD in relevant fields such as robotics or AI, exhibit strong programming skills, and have a robust publication record. Applications are accepted until August 3, 2025, with inquiries directed to Prof. Ville Kyrki. (Source)

Green-collar job demand has surged to an eight-year high, with an average of over 21,000 monthly vacancies from January to April, according to a report by the Ministry of Environment. The high-tech industry leads this demand, accounting for 22 percent of openings, largely due to its scale and rigorous regulations. Environmental health and safety (EHS) roles comprise 15 percent, or over 3,300 vacancies, emphasizing the need for specialists to aid in legal compliance with net-zero goals. Notably, the fusion of green-collar jobs with artificial intelligence expertise has seen a 398 percent increase in demand, now making up nearly 20 percent of such vacancies. Minister of Environment Peng Chi-ming highlighted the necessity for workforce training, as industries increasingly incorporate AI to achieve net-zero transition by applying existing technologies. (Source)