Nvidia: The Most Valuable Enterprise Gen AI Use Cases
Over the last year, generative AI (Gen AI) has revolutionised the corporate world at an unprecedented pace. Following the introduction of ChatGPT by OpenAI just over 12 months ago, a study by the IBM Institute for Business Value has revealed that 50% of CEOs globally are already embedding Gen AI into their digital offerings. Furthermore, 75% of these leaders are convinced that businesses at the forefront of Gen AI technology will secure a competitive edge.
In contrast to conventional AI, which is designed to identify patterns and facilitate decision-making, Gen AI excels in creating new data from the patterns it discerns in existing datasets.
Research from MIT underscores the potential, showing that companies could see productivity surge by 50% through the adoption of Gen AI. The same study noted an uptick in job satisfaction and performance among employees who utilise AI tools.
In today’s world, there are a fast-growing number of Gen AI use cases, including language, image, audio and video generation. Here, Erik Pounds, Senior Director of Enterprise AI at Nvidia, explores three popular ones in more depth – intelligent search, virtual assistants and summarisation, all of which are set to power a wave of advancements in the year ahead.
Virtual assistants boosting customer experience
Businesses are continuously exploring innovative ways to enhance their overall efficiency and effectiveness. A key focus area is improving productivity by streamlining internal operations, as well as elevating the customer experience to drive satisfaction and loyalty.
One increasingly popular solution, Pounds highlights, that businesses are turning to in order to achieve these goals is the deployment of AI-powered chatbots, virtual assistants and intelligent copilots.
“Enterprise application developers have started to use a customisation technique called retrieval-augmented generation (RAG) to connect Gen AI models with their enterprise data to deliver highly accurate responses for their AI-powered applications,” he says.
“Connecting LLMs to multiple data sources and knowledge bases lets users easily get up-to-date answers with the knowledge of their data using simple, conversational prompts. With the option of adding RAG capabilities to applications, developers can create customised virtual assistants, chatbots or copilots that can engage in an interactive conversation with users both internal and external to the organisation.”
Beyond enhancing the customer experience, the deployment of virtual assistants and other AI-powered technologies can also yield significant benefits for a business's internal operations and employee productivity.
When utilised within the organization, these intelligent digital assistants can serve as powerful personal helpers for individual employees. They can provide tailored support across a wide range of job functions and responsibilities, empowering workers to be more efficient and effective in their day-to-day tasks.
“Imagine having a virtual assistant that can help with financial analysis, create assets for marketing campaigns or provide real-time sales insights,” he adds. “These intelligent companions empower employees to be more efficient and effective in their roles, freeing up time for higher-value tasks.
“Moreover, virtual assistants are being used to discover new or better revenue generation opportunities, from basic product inquiries to complex purchasing decisions. Through natural language understanding, chatbots can comprehend and respond to customer queries in a conversational and human-like manner, enhancing the overall customer experience.”
Leveraging intelligent search
Intelligent search, long a staple of the internet, has become a part of the daily lives of billions of people worldwide. This process, Pounds explains, is powered by LLMs trained on internet-scale datasets to give it broad knowledge of the natural languages and skills required to understand the intent behind user queries.
“There is a wealth of proprietary data available within enterprises in the form of private documents, databases, and mission-critical applications on platforms such as Snowflake Data Cloud or Oracle Cloud ERP, which are integral to business operations,” he says. “The knowledge of the business or organisation is contained in this proprietary data, but leveraging it to its full potential has been a challenge.
“With Gen AI, enterprises start with a generic LLM, sometimes referred to as a foundation model, that has been pre-trained on vast amounts of publicly available data so that it understands human languages and has a broad range of general knowledge.
“Once a foundation model is customised with proprietary business data, enterprises can create applications that understand a company’s specific lingo and deliver up-to-date, business-specific search results for both employees and customers. In many use cases, a second LLM acts as a monitor for the primary LLM to provide guardrails that prevent interactions from veering into unwanted territory.”
Gen AI for summarisation
Distilling lengthy documents or virtual meetings into concise, easy-to-digest notes and bullet points has long been a pain point for businesses. Traditional approaches often require manual effort, which can be time-consuming and prone to errors.
However, the rise of generative AI models is transforming this challenge. These advanced AI systems can now analyze the contents of documents, recordings, or videos, and automatically generate concise summaries within seconds. By leveraging their natural language understanding capabilities, they can break down complex content and convey the key information in plain, accessible language.
“Over the next year, summarisation capabilities will come to many of the productivity platforms that individuals rely on and will increase day-to-day efficiency,” Pounds describes. “These models can help internal teams across industries. Doctors and medical professionals, for example, can leverage Gen AI to summarise patient notes, enabling them to quickly grasp crucial information and deliver better care.
“Researchers at NYU Langone Health, the academic medical centre of New York University, are developing an LLM fine-tuned with 10 years’ worth of a patient’s health records to predict that patient’s risk of 30-day readmission, as well as other clinical outcomes.
“In the financial services industry, AI models analyse thousands of data streams in real-time, gleaning insights from market intelligence to create summarised research reports and deliver improved investment returns for investors and portfolio managers.”
A complete end-to-end platform for Gen AI
To get started with adding Gen AI capabilities to enterprise applications, Pounds explains how it is critical to have an AI foundry to create custom Gen AI models and an enterprise-grade platform to run them in production anywhere.
“A full-stack platform includes accelerated infrastructure, essential software that powers the end-to-end AI workflow, Gen AI tools, pre-trained foundation models that can be easily customised and solutions that connect in proprietary business data using RAG.
“No business has to make the journey to Gen AI alone,” he concludes. “Industry experts and consulting leaders are building out practices to guide enterprises through adding Gen AI capabilities to their businesses. The industry will continue to evolve at a rapid pace, and enterprises with strong partnerships will be well positioned to adapt their strategies and lead their industry in Gen AI.”
******
Make sure you check out the latest edition of Technology Magazine and also sign up to our global conference series - Tech & AI LIVE 2024
******
Technology Magazine is a BizClik brand
