Morten Rand-Hendriksen’s Post

A fascinating example of the power of AI and the need for a human in the mix. Yesterday AI and education expert Ethan Mollick shared a demo of Claude 3.5 building an artifact demonstrating Thomas Kuhn's theory of scientific revolutions. It's an impressive technological achievement, and what Claude produced is also fundamentally incorrect. This is a great reminder of the power of AI and the responsibility we have to ensure what the AIs tell us is accurate. Ethan's original post for reference: https://lnkd.in/g-kE8PyZ #ai #openai #claude #claude35 #artifacts #claudeartifacts #education

EU regulators label X (Twitter) for-pay blue check verified symbol deceptive design because that's what it is. "Since anyone can subscribe to obtain such a 'verified' status, it negatively affects users' ability to make free and informed decisions about the authenticity of the accounts and the content they interact with," I predicted this back in the fall of 2022 because it was immediately obvious this would eventually happen. Elon destroyed the verified system because he believed in an unhinged right-wing conspiracy theory about the system being used to silence conservative voices. In response he created a for-pay system that actively misleads users.

RAG, fine-tuning, RAFT, vectors, and knowledge graphs are all converging in the current endeavor to make #GenAI work in the enterprise. In this short #LinkedInLearning course I explain what these things are all about so you know what you're talking about and where the AI conversation is currently headed! Check out this video! I have made it free for you with the link below. #ai #rag #finetuning #vectors #vectordatabase #vectordb #graphrag #graft #raft #llm #artificialintelligence #enterpriseai #knowledgegraphs

🐙 The octopus offers a useful mental model for understanding AI systems and agents. Each tentacle of the octopus contains its own neural network, functioning as a task-based brain system. When the octopus perceives something like a crab, it signals a tentacle to take action, invoking a specific program. The evolving interaction between local Small Language Models (SLMs) and cloud-based Large Language Models (LLMs) highlights an interesting dynamic in AI development. When envisioning an enterprise agent today, a mix of specialized fine-tuned SLMs and LLMs, guided by a central orchestrator similar to what we see in the octopus is an approach worth exploring. The orchestrator determines the most suitable model (tentacle) for the task at hand and the appropriate program to execute, leaving the heavy specialized lifting to run autonomously. Rather than relying on a single big and costly LLM for all functions, leverage cost-effective SLMs for most tasks and turn to LLMs only when their more powerful capabilities are required. This approach has a name: Federated Language Models. I expect you'll hear a lot about them in the future. Be like the octopus. -- Follow Morten Rand-Hendriksen for practical insights and critical analysis of all things AI. #ai #llm #slm #federatedlanguagemodels #octopus