Key Development Points of AI Large Models in 2024


The AI healthcare industry reaches a critical turning point as iFlytek Medical launches Spark Medical Large Model V3.5 on June 9, trained on domestic computing power. The model shifts focus from parameter scale to core scenarios of clinical diagnosis and resident health management, demonstrating practical clinical application through data from top-tier hospitals.....
In support of the 'Clear and Bright - Rectification of AI Application Chaos' special action, the Central Cyberspace Administration Office's Reporting Center has launched an AI application chaos reporting section this month, opening up reporting channels for the public. The section accepts 14 categories of violations, divided into two major categories: AI application service violations (such as unregistered large models, insufficient security reviews, risks in training data, data poisoning, and content fabrication) and AI-generated content violations, aiming to standardize AI product services and protect online users' rights.
At the opening of the 2026 US-Mexico-Canada World Cup, the Tongyi Qianwen App launched a football prediction AI assistant on June 12th. The assistant accurately predicted the opening match, Mexico 2:0 victory over South Africa and the red card trends on the first day, and also predicted South Korea's 2:1 win against Czech Republic, which was called the AI version of the octopus by netizens. The assistant is based on massive data training and introduces innovative factors such as the host country, demonstrating high accuracy.
Perplexity jointly released a report with the Harvard Business School, comparing the Perplexity Computer general AI agent with traditional search assistants. Traditional assistants only answer questions and require users to perform follow-up actions manually, while AI agents can autonomously plan, execute tasks, and produce results. Data shows that Perplexity Computer's AI agent runs autonomously for an average of 26 minutes per session, far exceeding traditional search assistants, demonstrating the comprehensive transformation of knowledge work by AI agents.
The AI programming tool Claude Fable5 demonstrated amazing autonomy. Technical blogger Simon Willison sent only a screenshot and a brief instruction "check dependencies and find the problem," and the AI independently fixed the page scrollbar bug in Datasette Agent, astonishingly showcasing its independent execution capabilities.