Ai for Business

  I’m quite fascinated by the human ability to learn language. In my recent article I had discussed how experiments are being conducted and AI models are being trained to understand early language acquisition in humans and whether machines can mimic it. I’m a neuroscience enthusiast and keep reading about neuroplasticity, our brain’s ability to change and adapt to our environment. Neuroscientists believe that neuroplasticity plays an important role in preventing cognitive decline and reduces the risk of dementia. Researchers say that one factor that promotes neuroplasticity is learning anything new, say for example, learning a new language. I’m myself learning Spanish and have been using the language learning app Duolingo for some time now! My interests in the fields of neuroscience and AI always want me to draw a parallel between the two fields and understand how one takes inspiration from the other. In the last article we saw how AI takes inspiration from neuroscience in understa...

I read 📖 a lot of posts on LinkedIn and Medium about LLMs. I distilled the top 5 topics of discussions  from my notes of the last two months. Here are the 5 topics. 1️⃣ RAG (Retrieval-Augmented Generation) techniques and improvements: Research topics include comparing RAG with fine-tuning methods, exploring advanced RAG techniques, multi-hop and agentic RAG architectures, and ways to improve the performance of retrieval-augmented models. 2️⃣ Use of LLMs in enterprise and government: This covers the application of generative AI within city governments, potential use cases for generative AI in organizations, ethics and governance in AI health applications, and the state of generative AI in enterprise settings. 3️⃣ Embeddings and their role in enhancing retrieval: This includes innovative techniques such as Matryoshka Embeddings, improving retrieval mechanisms through strategic inclusion of relevant and noisy documents, and blending models for improved performance. 4️⃣ Agents and...

  If you are implementing RAG in your organization, but can’t think of many ways in which it can be useful, read on. 🤔🤔 As everyone knows, RAG is the most popular way of consuming GenAI in enterprises. It is easy to get started, but hard to follow through in production. One of the reasons is that you need genuine use cases that will add value to your users. The more such use cases you find, the more traction your solution will gain. ⭐ Let’s see how to find these use cases. 1️⃣ The first thing to do is to find the long and complex documents in your organization. There are some such documents that every organization has- contracts, RFPs, proposals and so on. 2️⃣ Then there are those documents that are specific to your organization or your domain. Here are some examples: 👉 Standards and specifications: your team designs a solution that has to comply with certain standards. Or they may have to refer to specifications of materials and components. 👉 Compliance notifications: R...

❓Why do the users of RAG describe the answers as ‘correct but not useful’? How can the RAG pipeline be improved to avoid this? 😒 RAG user: ‘this answer is good, but it is not useful to me. It should have also mentioned a few more things’. 🤔 RAG Developer: ‘this looks perfectly good to me. I am myself amazed how my program is able to produce such a beautiful answer’. One will encounter many such interactions while putting RAG into production. During the prototyping, when RAG starts producing correct answers, there is a feeling of awe and surprise. But when the actual users start using it for day-to-day use, they find the answers very basic. They say that they can find these answers reasonably quickly without the tool. Why does this happen? 🤨 In my previous posts, I described how RAG works. It basically has three parts: 1️⃣Retrieve the material relevant to users' query. 2️⃣Augment the prompt with the material. 3️⃣Generate the answer using an LLM. Each of the above parts can be im...