Adapting and accepting GenAI in KM and content delivery

The pulse of business today necessitates fast, accurate access to unstructured content in order to thrive among the competition. Despite this being the end goal for enterprises, many struggle to deliver; end users are continuously faced with a seemingly endless search for information, reflecting poorly on an enterprise’s state of business and reputation.

David Seuss, CEO at Northern Light, and Johnny Kinnaird, director of AI success at Shelf, joined KMWorld’s webinar, Using Generative AI for Fast, Accurate Knowledge from Unstructured Content, to highlight the role that generative AI (GenAI) can play in optimizing unstructured content access, driving better decision making, lower operational costs, and increased revenue overall.

The heavy push for implementing GenAI has certainly prompted many professionals to question its efficacy for enhancing various workflows. According to Seuss and a report from Accenture, the pressure to adopt GenAI is coming from the top, where executive roles see value in using GenAI to enable connections across data types.

Despite GenAI’s promise for efficiency, many large enterprises—such as Apple, Amazon, and Samsung—have banned the use of GenAI models like the ever-popular ChatGPT.

Why are companies banning GenAI when it stands to offer proprietary gains? According to Seuss, inaccurate and unreliable information, as well as confidentiality of internal information, has led to enterprises prohibiting ChatGPT in the workplace. Misleading information from a GenAI hallucination or unauthorized access to sensitive data means GenAI can pose more harm than good to any business.

The fundamental problem with GenAI such as ChatGPT for a business context is that it is trained on public information, pulling content from a universe of information that can be misleading or untrue.

To combat this challenge, Seuss emphasized that any GenAI model adapted to a proprietary context should highly empower accuracy and privacy. He advised that GenAI models should:

• Provide citations so the user can vet the source and consume documents in-depth if desired
• Prompt the user to click through and verify the results
• Be sourced from a major GenAI player, such as OpenAI or Microsoft, that follows industry best practices for data security
• Perform the interaction with the LLM via a private enterprise API channel
• Anonymize the source of the content and users’ questions
• Implement “Zero Day Retention,” which means that the LLM does not store the document text or user question once the transaction is complete

Ultimately, Seuss concluded that GenAI has fundamentally changed the search paradigm, and the genie can’t be put back in the bottle. Every organization should prepare for a GenAI implementation, with accuracy and security at the forefront of the conversation.

Kinnaird backdropped his conversation of GenAI for unstructured content access with some historical context, explaining that the beginning of the technological revolution—thousands of years ago—was the beginning augmentation of the human ability to think, create, build, communicate, and manifest.

While it’s difficult to imagine how the advent of writing has lent itself to modern day AI, the steep acceleration of technological evolution is critical in understanding (and accepting) the rapid change of AI in the past year alone. Echoing Seuss, Kinnaird added that this era—as well as its rapid tendency to change—is not going anywhere.

“This is not a hype curve, but rather a fundamental shift in human civilization,” said Kinnaird.

With this shift representing such a fundamental and widespread change, it is felt everywhere—even in KM. With some fearing replacement by AI in the workforce, Kinnaird assured webinar viewers that knowledge managers have crucial core skills that will be needed to navigate AI, including:

• Mastery in the classification of knowledge
• Taxonomy and ontology design
• Mastery of English language for prompt engineering
• Research skills
• Editorial skills and supervision
• Information governance delivering oversight and quality assurance

While, “of course, there will be some areas that become automated to fully automated,” explained Kinnaird, “The role of KM professionals will only be further empowered by AI, not displaced. AI enables knowledge managers to do more, faster and learn more, better, ultimately amplifying their KM impact.”

Kinnaird then offered this list of advice for KMers in the age of AI:

• Learn as much as you can.
• Embrace the change and adapt.
• Use the new technology to tap into new potential.
• Reimagine your work.
• Actively create the future you want to live in.

For an in-depth discussion regarding AI and managing content access, featuring demos, examples, and a Q&A, you can watch an archived version of the webinar here.