Interesting Stuff - Week 17, 2024

Posted by nielsb on Sunday, April 28, 2024

This week on the blog, we dive into the transformative power of Causal AI, which is reshaping decision-making by focusing on causation rather than mere correlation. We also explore the realm of serverless AI chatbots built using LangChain.js and Azure, highlighting their scalability and efficiency.

Additionally, we discuss how graph-based metadata filtering is refining vector search in AI applications. Don’t miss our vibrant recap of the Data & AI Community Day, a testament to the thriving interest in these advanced technologies.


  • Understanding Causal AI: Bridging the Gap Between Correlation and Causation. This article from MarkTechPost explores the emerging field of Causal AI, which seeks to move beyond traditional AI’s focus on correlation to understanding causation—essentially, the “why” behind data. This approach integrates causal inference into AI algorithms, allowing for more sophisticated reasoning about cause-and-effect relationships. These capabilities are particularly beneficial in areas like healthcare, where understanding the impact of treatments on patient outcomes is crucial, and in economics, for assessing policy effects. The transition to causal reasoning could significantly enhance decision-making processes, making AI systems more robust and ethically sound by enabling them to make decisions based on clear causal relationships rather than mere correlations.
  • Build a serverless AI chat with RAG using LangChain.js. The blog post dives into constructing a serverless AI chatbot using LangChain.js, Ollama, and Azure, explicitly utilizing the Retrieval-Augmented Generation (RAG) model. It outlines a step-by-step guide for building a chatbot that leverages AI to answer questions based on retrieved information, highlighting the integration of various AI models and Azure services for deployment. The post emphasizes the efficiency and scalability of using serverless technologies, making it an excellent resource for developers interested in AI and cloud infrastructure.
  • Graph-based metadata filtering for improving vector search in RAG applications. This blog post on LangChain discusses enhancing vector search in Retrieval-Augmented Generation (RAG) applications through graph-based metadata filtering. It explores how combining structured graph data with unstructured data can refine search results by filtering based on metadata like dates, categories, or relationships within data. The method employs LangChain and Neo4j to create a more targeted and effective retrieval process. This approach improves the precision of search results in complex datasets by incorporating structured and connected graph data into the search mechanism.

WIND (What Is Niels Doing)

It’s a Wrap!

Figure 1: Data & AI Community Day April Durban

Yesterday marked an extraordinary gathering of minds at the Data & AI Community Day Durban, held in the stunning offices of BETSoftware in Umhlanga. As we ventured into this event, our expectations were high, but the reality far surpassed what we had imagined. The turnout was incredible, with many more attendees than anticipated, highlighting the growing interest and enthusiasm in the fields of data science and artificial intelligence.

Figure 2: Packed Room

From the onset, the energy in the room was palpable. Each session was packed, with several talks reaching standing-room-only capacity—a testament to the quality and relevance of the presentations delivered. It was inspiring to see such a diverse group of participants, ranging from industry professionals to academic enthusiasts, all eager to share insights and learn from one another.

The feedback from the attendees has been overwhelmingly positive. Conversations buzzed with excitement and appreciation, and the evaluation forms reflected a high level of satisfaction with both the content presented and the organization of the day. This success would not have been possible without the immense support of our sponsors and speakers, whose contributions were pivotal. Their commitment to fostering a knowledgeable community is deeply appreciated.

As the day drew to a close and the last of our attendees departed, I found myself overwhelmed by the whirlwind of organizing and running such a large-scale event. In that moment, I swore to myself that this would be my last.

However, after a restful night’s sleep, my perspective shifted dramatically. Refreshed and reinvigorated, I am now more eager than ever to dive back into the fray. The success and impact of this event have inspired me to start planning the next one, slated for July. Stay tuned for more details as we hope to bring even greater experiences to our community.

~ Finally

That’s all for this week. I hope you enjoy what I did put together. Please comment on this post or ping me if you have ideas for what to cover.

comments powered by Disqus