Engineers have made great strides in recent decades creating software capable of processing information with the same efficiency of the human brain. The pursuit of human-like Artificial Intelligence (AI) was the central theme of an on-stage conversation at this year’s Digital Workforce Summit (DWS) between IPsoft CEO Chetan Dube and Stanford University Professor Christopher Manning.
Manning is an instrumental partner in IPsoft’s continuing development of its leading virtual agent Amelia, who is routinely recognized by industry analysts as the most advanced conversational solution on the market.
The two AI luminaries had a wide-ranging and fascinating discussion that explored how AI has evolved over time, particularly in its ability to engage with humans through natural language. “The last five years have been an extremely exciting period in the development of natural language processing technology,” Professor Manning said.
“I’ve been involved in natural language processing for 25 to 30 years now,” he said. “Over the last five years, we’ve actually seen just a super rapid period of improvement where we’re just doing a much better job at building models that can understand language in its context of use and we’ve been able to build some of that technology into Amelia.”
Over the last five years, we’ve actually seen just a super rapid period of improvement where we’re just doing a much better job at building models that can understand language in its context of use and we’ve been able to build some of that technology into Amelia.
— Christopher Manning, Professor, Stanford University
Unlike low-level solutions like chatbots which follow static rules, Amelia provides a far more robust user experience by organically placing interactions in a grander context – just as the human brain does. “Human language connects indirectly to our higher-level cognitive processes,” Manning said. “Cognitive processes are where we reason and plan, where we understand requirements for something to happen, and the consequences for taking one course of action. And so we need to model these higher-level human reasoning [capabilities] to understand complex scenarios and then express [them in] natural language.”
The key, Manning said, is to meld together logical with statistical and empirical Machine Learning capabilities. “I believe we’ll only reach the true power of Artificial Intelligence when we can actually span across both ends of that continuum,” he said.
Bridging the Divide
Professor Manning’s work with IPsoft exists at the crossroads of philosophy and engineering. His insights into the way the human mind functions assist IPsoft engineers with replicating advanced functions in Amelia. “Humans are very good at maintaining context, understanding changes, working out plans to process these things – and it’s precisely that kind of richness of contextual modeling that is our goal in Amelia,” Manning said.
Of course, for all of Amelia’s accomplishments, she remains an evolving solution that is constantly updated and improved. When asked by Chetan what the future holds for “this little child of ours of the last five years,” Professor Manning reiterated Amelia’s accomplishments before transitioning into the “challenges of understanding.”
“They’re challenges of understanding the world. Because to really understand the implications of things that are said, you have to understand people’s lives and how people’s worlds work and that’s something we’re trying to work on,” Manning concluded. “The world is a very complex place and having that higher level of cognition and understanding is still somewhere where there’s quite a lot of work to do.”
For additional DWS perspectives and media coverage, head over to our IPsoft Newsroom.