KMWorld 2024 Is Nov. 18-21 in Washington, DC. Register now for $100 off!

Navigating the challenges and advantages of generative AI for KM

Winning the attention of millions, the promises of generative AI inspire innovators in every industry. Knowledge managers are no different than the innovators leaping onboard the generative AI train, as its potential to revolutionize KM—whether through dynamic content generation, discovery automation, or surfacing greater data potential—is certainly enticing. However, like any other trend in technology, generative AI still poses unique difficulties that can impede its success.

Experts in KM joined KMWorld’s webinar, Recent Trends in Generative AI, to explore the technological phenomenon’s potentials and drawbacks, offering ways in which AI can be leveraged to invite transformative, modern KM.

Keith Berg, SVP and general manager of contact center solutions at Upland Software, highlighted that while knowledge management is made up of many components, content management can be the most time consuming. In the same breath, Berg explained that though generative AI seems so new and exciting, enterprises have always used AI and machine learning in authoring tools; generative AI just supercharges the experience.

Berg then clarified that with Upland AI Knowledge Assistant, the future of KM is merely a click away. The solution can:

  • Create knowledge through direct query or add ticket information to auto create problem solving articles
  • Improve content quality through grammatical auto-check
  • Simplify the structure and flow of troubleshooting information with sequential instructions or intuitively categorized information
  • Auto summarize to simplify answer search

Despite the expansive advantages of AI in KM, Berg cautioned that it isn’t ready for everything. This includes eliminating human approval of answers, replacing knowledge managers or subject matter experts, and automating review and approvals.

David Seuss, CEO at Northern Light, dove into a thorough explanation of the popularity of ChatGPT, as well as the differences between ChatGPT and large language models (LLMs). Ultimately, he directed viewers to a poignant statement: question-answering is at the heart of the business use case for generative AI.

Tackling this use case, Seuss pointed to Northern Light’s SinglePoint Platform, a custom-built enterprise knowledge management platform that seamlessly integrates and enables full-text search of all of an enterprise’s unique research resources. The platform provides:

  • Generative AI summaries base on relevant documents that aim to answer user queries
  • Citations that enable users to vet their generated information’s source
  • Links to documents for in-depth research
  • Support for a myriad of content types, including business and technology news, market intelligence, industry analysis reports, technical journal articles, and more

The platform utilizes GPT-3.5 Turbo from OpenAI, which produces quality answers when used specifically on business and technical content, according to Seuss. Regarding security, Seuss referred to OpenAI’s contractual tenets; the enterprise agrees not to store user’s text or questions, not to access content for any purpose, and isolate all search transactions from other search transactions.

Bonnie Chase, senior director of product and content marketing at Coveo, explained that due to the advent of generative AI, what people expect has changed for good. They believe that content should be tailored to the individual, that each interaction should be used as a learning tool to inform the next, and that each journey should be unified—not channel-specific—in nature.

Although generative AI has the capacity to deliver on all these fronts, it still incurs a variety of headaches that must be dealt with on the enterprise side. This includes:

  • Security of generative content
  • Privacy of public generative engines
  • Protecting enterprise proprietary content
  • Dealing with multiple content sources and the currency of content
  • Applying factuality/veracity at scale
  • Ensuring coherence of search and chat channels
  • Providing sources of truth and verifiability
  • Confronting the high costs of generative AI

To address these abundant challenges, Chase recommended the Coveo Relevance Generative Answering platform, an advanced generative AI solution that combines accurate and relevant question-answering with enterprise-grade security. The solution utilizes LLMs in conjunction with Coveo's secure and powerful Relevance Cloud AI platform to bring users closer to where their answers lie.

Coveo’s AI-powered search platform offers streamlined and personalized navigation and answer-response designed with relevance at the forefront. Users input a query with their natural language, and the platform uses LLMs to generate an answer.

The backend of the solution contains content of breadth, depth, and freshness, protected with robust security and manageable through administrative capabilities and analytics. On the results and relevance end, users experience a unified search box for all queries that generates answers that are both personalized and protected against hallucination.

For an in-depth review of generative AI challenges and solutions, you can view an archived version of the webinar here.

KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Companies and Suppliers Mentioned