-->

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

  • November 4, 2024
  • By Marydee Ojala Editor in Chief, KMWorld, Conference Program Director, Information Today, Inc.
  • Features

Lucidworks shares results from a global survey about GenAI adoption, fears, and strategies in 2024

Article Featured Image

Lucidworks recently published its second annual global generative AI (GenAI) study, titled “Generative AI and the B2B Marketplace.” Mike Sinoway, Lucidworks CEO, spoke with me about the survey results and his observations on the effect of GenAI on knowledge management. This article is a summary of the key insights; the full podcast is here: kmworld.com/Podcasts/13672-5-Takeaways-From-Lucidworks- Second-Annual-Global-Gen-AI-Study.htmkmworld.com/Podcasts/13672-5-Takeaways-From-Lucidworks-Second-Annual-Global-Gen-AI-Study.htm.

Sinoway defines GenAI as it applies to the business environment as an extension of how AI and machine learning have been used for decades, which had been primarily numeric. GenAI now allows for the same applications, but with text, video, and audio. The same tools we applied to numbers we’re now applying to words. GenAI analyzes literally millions, even billions, of data-points and makes predictions about the next letter in a word, the next word of the sentence, or the next sentence in a paragraph. It then comes up with a response to answer a query. Its predictive qualities usually produce good answers, but the possibility of hallucinations, which is what incorrect predictions resulting in wrong answers are called, is worrying.

Declining hype

When Lucidworks did its first survey in 2023, the hype around GenAI was red hot. That has cooled off, as the second survey shows. Realism is setting in—these tools won’t deliver everything people expect. A year ago, 93% of the companies in the study were planning to increase spending on GenAI. A year later, it’s down to 63%, which is still a large number, and it varies considerably by country. The use cases cited in the first survey were focused on revenue-generating activities, but only 1 in 8 of those initiatives actually came to fruition. In contrast, 1 in 3 KM projects succeeded.

Sinoway noted that companies are more attuned now to the possible risks in implementing GenAI. They’re concerned about it making a mistake and hallucinating a wrong answer. They also worry about data security. For external applications, it’s highly visible if something goes wrong. KM use cases are generally internal, where there’s concern but not as much hesitation. If you give a slightly wrong bit of information to another employee in the company, usually you have a chance to find that and correct it before it affects the organization or the customers. Lucidworks sees KM use cases being implemented much more frequently because of that lower risk potential.

Building a business case

Investing in GenAI requires building a business case to justify the investment. What is the business benefit, the ROI, from using the technology? What tool and what data are needed? For external use cases, the business benefit is revenue. Internally, it’s productivity. Lucidworks’ second survey revealed that companies had moved from thinking GenAI would drastically reduce head count to realizing it’s not about replacing people, it’s about making them more productive.

According to Sinoway, there are several pieces to the AI puzzle, of which GenAI is just one piece. Cost-effective solutions for business problems is another. There has been an evolution in how AI is being deployed in companies. RAG (retrieval-augmented generation) is being used to cut down on the hallucination problem. Instead of using large language models (LLMs) that index all the information in the world, RAG uses an LLM to index internal information about customers, products, and costs; to augment that with trusted external data; and to use GenAI to pull accurate data. When LLMs use internal data, they’re smarter and they’re cheaper.

KMWorld Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues