New research from Google and MIT shows multi‑agent AI often drops context in sequential tasks, unlike single‑agent models. The study highlights token‑budget limits and proposes fresh benchmark suites for web‑search style challenges. What does this mean for open‑source AI development? Dive into the findings. #GoogleAI #MIT #MultiAgentAI #Benchmark

🔗 https://aidailypost.com/news/google-mit-study-finds-multiagent-ai-often-loses-context-sequential