AI Dinner 11.0 — The Recursive Illusion Of Thinking

Thanks to Arize for sponsoring the space this month!

This month is packed with strong updates and new papers. Please read the AI report and dive into 2–3 key topics or papers. It will help you contribute meaningfully and keep our Socratic discussions flowing. Agenda 6:30 PM Dinner starts 7:00 PM Socratic Conversations:Intro around the room and announcements Socratic conversations on the latest in AI from June 2025, new models, and research papers: https://site.flowai.xyz/ai-socratic-june-2025Anthropic multi-agent advanced search Mistral Magistral GRPO edits The Illusion of Thinking and The Illusion of the Illusion of Thinking (first paper authored by an LLM) Learning to Reason Across Parallel Samples for LLM Reasoning Jianing Qi Demos & PresentationCommenting on chess moves in natural language (training LLM via RL to play chess) Kerem Kazan DM to suggest a 5-15 min presentation: https://docs.google.com/document/d/1UEbZP1UqCCDyLtfm2Y-ZnuYzLRNPwK-odXc6oM8SPp4 9:00 PM Networking

The AI NYC 🗽🍎 is a community of top-tier applied AI researchers, engineers, and founders meeting regularly to cut through the noise of the AI news cycle, bridging the signals across all verticals. Some of our past discussions include: MCP Workshop: we focused the entire event on learning how MCP works and showing demos. DeepSeek Study Group, we read the R1 and V3 papers and went into the rabbit hole of reasoners and show how to install DeepSeek. SOTA RAG: integrating search and retrieval mechanisms like HyDE with LLMs for enhanced knowledge retrieval in real-time applications.

Sponsors

Thanks Arize and Aman for sponsoring the office space!

Thanks to CoopHive for co-sponsoring the event food.

Thanks to Anissa (CryptoCouture) for preparing the middle eastern inspired dinner and for curating vibes.

Fed (Flow), Ivo (Trebuchet) are moderating the event and co-sponsoring food and drinks:

👉 Important The AI NYC community is exclusive to applied AI researchers, engineers, and experts focusing on AI/ML. We review every application, so make sure you go into the details of what you're working on and what value you bring to the conversations.