Generative AI is transforming knowledge work across industries. Companies are adopting generative AI tools and establishing policies for their safe use. Individuals can also benefit from using generative AI to reduce cognitive load and enhance learning. It's time to start using generative AI wisely.

The article discusses the challenges of implementing "Chat over Your Data" and explains why it is more difficult than it may seem. It provides insights into the complexities involved in building chat functionality that operates directly on user data and highlights the importance of data privacy and security. The article is available at the following link: https://www.arcus.co/blog/chat [https://www.arcus.co/blog/chat].

Interest in large language models, such as those developed by ChatGPT maker OpenAI, has put renewed focus on data management—placing more pressure on corporate technology chiefs to ensure their companies’ data is stored, filtered, and protected for use with AI.

The article provides guidance for Chief Data Officers (CDOs) on using data to fuel generative AI. It emphasizes the importance of focusing on value, building specific capabilities in the data architecture, ensuring data quality, protecting sensitive data, developing data engineering talent, and tracking progress and metrics.

The integration of knowledge graphs with generative AI enables trustworthy and responsible enterprise decisions. Knowledge graphs provide a structured representation of knowledge, while generative AI can produce new content based on patterns in its training data. This whitepaper explores generative AI use cases, potential pitfalls, and how knowledge graphs can mitigate risks. It also discusses empowering organizations with knowledge-enriched generative AI and a comprehensive approach to its construction.

Originally posted on The Horizons Tracker. Few AI-based tools have made such a splash as that made b

Large language models (LLMs), such as ChatGPT and GPT4, are making new waves in the field of natural language processing and artificial intelligence, due to their emergent ability and generalizability. However, LLMs are black-box models, which often fall short of capturing and accessing factual knowledge. In contrast, Knowledge Graphs (KGs), Wikipedia and Huapu for example, are structured knowledge models that explicitly store rich factual knowledge. KGs can enhance LLMs by providing external knowledge for inference and interpretability. Meanwhile, KGs are difficult to construct and evolving by nature, which challenges the existing methods in KGs to generate new facts and represent unseen knowledge. Therefore, it is complementary to unify LLMs and KGs together and simultaneously leverage their advantages. In this article, we present a forward-looking roadmap for the unification of LLMs and KGs. Our roadmap consists of three general frameworks, namely, 1) KG-enhanced LLMs, which incorporate KGs during the pre-training and inference phases of LLMs, or for the purpose of enhancing understanding of the knowledge learned by LLMs; 2) LLM-augmented KGs, that leverage LLMs for different KG tasks such as embedding, completion, construction, graph-to-text generation, and question answering; and 3) Synergized LLMs + KGs, in which LLMs and KGs play equal roles and work in a mutually beneficial way to enhance both LLMs and KGs for bidirectional reasoning driven by both data and knowledge. We review and summarize existing efforts within these three frameworks in our roadmap and pinpoint their future research directions.

Retrieval-augmented generation (RAG) is a technique that enhances generative AI models by incorporating facts from external sources. It addresses the limitations of large language models (LLMs) by providing authoritative answers with citations. RAG connects generative AI services to external resources and can be used by any LLM to access a wide range of knowledge bases. It improves user trust, reduces ambiguity, and enables a variety of applications across industries. NVIDIA has developed a reference architecture for RAG and offers software components like NVIDIA NeMo and Triton Inference Server for running generative AI models. The technique has a rich history and is considered the future of generative AI.

The article explores the limitations of Large Language Models (LLMs) and presents two concepts to overcome them: fine-tuning and retrieval-augmented use of LLMs. Fine-tuning involves the supervised training phase, where question-answer pairs are provided to optimize the performance of the LLM. Retrieval-augmented generation uses the LLM as a natural language interface to access external information, thereby not relying only on its internal knowledge to produce answers.

This article is part 1 of the series AI integration strategy for learning and knowledge management s