KMWorld 2024 kicks off with a major focus on humans and AI working together

We need new frameworks for AI-powered decision making that keep humans in the loop (along with human values, morals, interests, emotions, and sensibilities).

At KMWorld 2024, Louis Rosenberg, CEO, Unanimous.AI and Author, Our Next Reality: How the AI-Powered Metaverse Will Reshape the World, opened the first day of the conference with his discussion on “Collective Superintelligence: Humans in the Loop.”

Rosenberg discussed an approach toward enabling collective superintelligence that is rooted in hundreds of millions of years of evolution, which is why it so greatly outperforms old-school methods that treat humans as mere datapoints to be aggregated.

Humans are not data. Humans are powerful data processors. The most viable pathway to collective superintelligence is to connect people together in real time and allow them to act, react, and interact using AI as the interstitial tissue that empowers us to solve problems together in optimal ways. He shared his insights and ideas for enterprises looking for ways to share knowledge in their organizations.

He explained that ASI or AGI are AI systems that can outperform the smartest humans across a vast range of cognitive tasks. The only unknown is how quickly it will be realized.

“If this happens, it would significantly change all aspects of society,” Rosenberg said. “Most AI researchers believe this will happen.”

LLMs are trained on large databases of information such as text documents, videos, and more. But will ASI solve all the complex problems across entire organizations? No, these systems lack knowledge of the organization, he said.

“I’m talking about connecting people into these systems to optimize productivity at superhuman levels,” Rosenberg said.

To build collective superintelligence, he explained, you start with swarm intelligence. It’s nature’s method for amplifying the intelligence of large groups. AI enables swam intelligence to collect information among networked human groups.

“These large-scale conversations work to make people smarter,” Rosenberg said.

Unanimous.AI worked on several academic studies to prove that Swarm AI worked. When humans worked together with this AI system, the accuracy increased exponentially. However, enabling large conversations with groups, the system buckles, he said. To leverage the knowledge and wisdom of large groups, Swarm intelligence can gather this information.

Knowledge Graphs and GenAI

GenAI retrieval-augmented generation (RAG) uses natural language understanding (NLU) and natural language generation (NLG) capabilities of large language models (LLMs) to securely support conversational search and discovery over enterprise content and data repositories.

But GenAI and RAG alone are not enough to ensure the completeness and accuracy of information for many mission-critical enterprise applications.

Knowledge graphs (KGs), including enterprise taxonomies and ontologies, can significantly improve the completeness and accuracy of information retrieved and generated by GenAI applications.

Dave Clarke, EVP, semantic graph technology, Synaptica, part of Squirro AG, U.K., discussed “Using Knowledge Graphs to Improve GenAI,” during his keynote.

“We help people to discover, classify, and organize knowledge,” Clarke said.

Taxonomies and ontologies provide GenAI with machine-intelligible context about the domain knowledge and processes of the enterprise. When KGs and GenAI are integrated, taxonomists and ontologists can see and rapidly edit graph structures that explicitly guide RAG decision-making processes.

With a simple no-code interface, taxonomists and ontologists are empowered to directly control GenAI dependencies, query refinement, and outcomes, thereby delivering high-quality, high-value business process automation.

“I love GenAI and it’s totally changing the world but, know its strengths, know its weaknesses…it can only get you so far,” Clarke said.

Knowledge graphs improve GenAI, its accuracy, insights, automation, and personalization. Squirro provides tools to enhance that, he said.

CX in the Age of GenAI

There is no question that GenAI has reignited interest in KM. Gartner predicts that 100% of GenAI virtual customer assistant and virtual agent assistant projects that lack integration to modern KM systems will fail to meet their CX and operational cost-reduction goals by 2025. As businesses experiment with GenAI, they are realizing that robust KM is foundational to its success.

Ashu Roy, chairman and CEO, eGain Corporation, discussed how KM and GenAI can accelerate and ensure mutual success, creating transformational business value at warp speed during his keynote, “Trusted knowledge for customer service in the age of GenAI.”

“I think we’re in the middle of a huge storm,” Roy said. “We are at the beginning of industrialization of knowledge work.”

If the data and information is garbage, the output will also be garbage, he explained. For GenAI to be helpful, what it’s fed needs to be correct. Disconnected silos and tools are making it difficult to get right answers.

“Every inconsistent input leads to incorrect answers from the system,” he said.