|
KDD2026-ReplicatorBench: Benchmarking LLM Agents for Replicability in Social and Behavioral Sciences
Association for Computing Machinery (ACM)
ReplicatorBench is a new benchmark that evaluates LLM agents on replicability studies in social and behavioral sciences, testing whether agents can validate whether research claims can be replicated using newly searched data — not just pre-provided datasets. The benchmark covers 39 studies with both replicable and non-replicable claims, features a three-stage process-oriented design with 3,000 evaluation checkpoints, and pairs each stage with manual documentation from human replication efforts. The authors also developed a baseline "Replicator Agent" using a reason-action agentic framework with a Docker sandbox and interactive debugging. Key findings show that while models excel at code generation, they struggle with collecting correct replication data from the web and with accurately interpreting final results.
|
|
How Students Are Making Money with AI Agents in 2026 | AI Voice Agent Crash Course
Saumya Singh
यह वीडियो एक क्रैश कोर्स है जिसमें बताया गया है कि स्टूडेंट्स AI Voice Agents बनाकर पैसे कैसे कमा सकते हैं। होस्ट सौम्या सिंह (एक सॉफ्टवेयर इंजीनियर) ने प्रैक्टिकल डेमो दिखाया जिसमें उन्होंने Akash Institute के लिए एक AI voice agent बनाया जो स्टूडेंट्स की कोचिंग इनक्वायरी, बैच टाइमिंग और एडमिशन संबंधी कॉल्स हैंडल कर सकता है। वीडियो में बताया गया कि लोकल बिज़नेस, स्टार्टअप्स और हेल्थकेयर जैसी इंडस्ट्रीज़ इन AI agents के लिए पैसे देने को तैयार हैं। साथ ही प्रो-टिप्स दिए गए — जैसे GitHub पर प्रोजेक्ट पोस्ट करना, real business problems solve करना, और LinkedIn पर अपना पोर्टफोलियो बनाना ताकि hiring managers की नज़र में आ सकें
|
|
agentic ai cringelords
Eric Morrison
The video is a satirical critique by Eric Morrison mocking the trend of "agentic AI hustlers" on social media — people who post viral videos of elaborate AI assistant setups (like a "Jarvis" giving them spoken business reports) to flex supposed autonomous AI-run businesses. Morrison argues these setups are likely staged, inefficient, and more expensive than simply checking a dashboard, pointing out that one creator's agent delivered the exact same report word-for-word in two separate videos. His broader point is that while AI agents are genuinely useful technology, a false and anxiety-inducing narrative has emerged — pushed by social media grifters — that you must deploy swarms of agents for everything or be left behind, and that people are actually buying into this hype to their detriment.
|
|
What Is Agentic AI? | Agentic AI Tutorial For Beginners | Agentic AI Explained | Simplilearn
Simplilearn
This Simplilearn tutorial explains agentic AI using the story of Emma, a small online clothing business owner, to contrast it with traditional AI chatbots. While normal AI only provides advice (e.g., "run discounts, post on social media"), agentic AI goes further by understanding a goal, creating a plan, and autonomously executing tasks such as writing emails, drafting social media captions, and building a campaign calendar. The video emphasizes that agentic AI operates step by step toward outcomes rather than just giving answers, but should always require human approval before taking significant actions like sending emails or spending money. It concludes by promoting Simplilearn's agentic AI certification as a way to build structured, professional skills in working with these systems.
|
|
LLM Wikis and how to give your agents memory
LangChain
This LangChain webinar brings together three guests — Brace (LangChain), Devon Stein (CEO of DOSU), and Jeff (Chroma) — to discuss LLM wikis and agent memory. Brace demos **OpenWiki**, a newly released CLI tool (installable via npm) that automatically ingests data from sources like Notion, Gmail, and Slack, then uses a nightly cron job to generate and maintain structured personal or codebase memory without manual upkeep. Devon demos **DOSU**, a knowledge infrastructure platform that plugs into coding agent workflows via MCP, using a "librarian agent" to capture learnings from agent tasks as concise, human-readable topic pages that are auto-injected into future sessions and kept up to date as code changes. Jeff from **Chroma** frames wikis as a new kind of index layered on top of unstructured data — analogous to adding a road network over a map — enabling agents to traverse knowledge more reliably than raw embeddings alone, while noting that Devin holds the "hot take" that wikis may be the wrong abstraction altogether.
|