By Arun Chandrasekaran
Generative artificial intelligence (AI), in isolation, is neither an alternative nor a replacement for current search technologies. Nevertheless, AI-driven search engines can help reshape how users interact with search capabilities through natural language processing (NLP) and machine learning (ML).
Gartner estimates that, by 2027, generative AI models will underpin 60% of NLP use cases, a major increase from fewer than 10% in 2022.
Revolutionizing the Future of Search with Generative AI
Generative AI can enable a broad range of search-related tasks, including the following:
- Summarization — Generating summaries to be returned in response to searches.
- Classification — Capturing and attributing metadata to digital assets and their parts.
- Query parsing — Reformulating user queries for processing against the index.
- Response quality — Checking the relevance of responses against the initial query.
Generative AI has the potential to transform enterprise search from a transactional to a conversational experience. AI foundational models, when used in conjunction with search technologies, can also power semantic, conversational search. Semantic search analyzes the relationships between words as well as words themselves during indexing and query to deliver more relevant search experiences. This has the potential to democratize access to information, where the users’ intent is better understood, and vital information is delivered to their fingertips. As a result, it enables them to make business decisions or perform tasks in a more productive manner.
Most businesses today are drowning in a massive sprawl of information, which is difficult to navigate and find. There are a variety of custom and bespoke tools that have been deployed with limited success. The potential to bring a conversational experience to organize and discover information, often in real-time, can be of enormous value for businesses across the globe.
Impact on Industries
Enterprise search is a horizontal use case and organizations in every industry will benefit from it. Knowledge industries such as information technology, life sciences, financial services, healthcare, telecommunications, media and entertainment will be some of the early adopters. For example, generative AI-powered enterprise search can enhance the clinician experience in healthcare, enabling care teams to access critical medical information at the point of care.
Deploying generative AI use cases also involve a level of sophistication in leveraging large language models with high degree of cognition, cohesiveness and knowledge grounding. There are a number of SaaS solutions emerging that promise to make enterprise deployments more seamless to implement.
Enabling Trust and Security
Security and governance are crucial for businesses before adopting any new tool or technology. To enable trust and security while using generative AI search, CIOs must take the following steps:
- Prioritize generative AI providers that provide data privacy guarantees around IP protection.
- Work with providers to deploy approaches such as prompt engineering and/or fine-tuning to reduce hallucinations.
- Choose providers that more closely integrate with enterprise data stores and can demonstrate robust access control, content rights and language understanding.
- Deploy solutions for explainability and content moderation to reduce harmful output and to demystify the black-box nature of these models. They should also conduct adversarial testing across a range of scenarios before deploying Generative AI apps in production.
- Create a mandatory training program for employees on the benefits and risks of Generative AI applications, including intended usage scenarios and how to provide user feedback.
Additional analysis on enterprise use of generative AI will be presented during Gartner IT Symposium/Xpo 2023, Kochi, November 28-30.
(The author is Arun Chandrasekaran, VP Analyst at Gartner, and the views expressed in this article are his own)
