IPsoft humanises artificial intelligence with the next generation of its cognitive agent

The growing maturity of Amelia’s core understanding capabilities will widen the range of roles she can execute and the breadth of knowledge she can absorb. In parallel, Amelia’s physical appearance and expressiveness have been transformed to create a more human like avatar capable of deepening customer engagement.  Her new animation form has been entirely remodeled on that of a real person.

Within a matter of years, IPsoft plans to put the model and her facesake, Amelia, on stage and have a distinguished panel of analysts and journalists ask questions of the human and the machine. “In just one year we have seen Amelia ‘grow up’ tremendously. Just imagine how her maturity will accelerate over the next five,” said Chetan Dube, IPsoft CEO.  “We are fast approaching the moment when technology is knocking at the Turing horizon, where machine intelligence starts to match human intelligence. Amelia will be the harbinger of that shift, inviting us to re-evaluate the relationship between man and machine in order to create a more efficient planet.”

As Amelia becomes more human in her interaction with customers, so the positive impact she has on business operations grows. Version 2.0 makes huge strides in terms of dialogue management comprehension and emotional engagement through the following technology developments:

  • Memory: Amelia’s declarative memory now consists of episodic memory and semantic memory, exactly the way human memory is organized. In humans, semantic memory activates the frontal and temporal cortexes, while episodic memory activity is mostly concentrated in the hippocampus. Similarly, episodic memory of Amelia gives her cognition of various experiences and events in time in a sequenced autobiographical form. Her semantic memory gives her a structured record of facts, meanings, concepts and knowledge about the client world.  The combination of the two allows Amelia to hold a wholly natural conversation that is not restricted to following set flows.
  • Contextual comprehension:Concepts and ideas in the human brain are semantically linked, so that thinking about or firing one set of neurons in your head primes other related ones, making them more likely to fire in future. This allows localized querying in the human brain as opposed to typical search based algorithms which traverse the entire data set before trying to compute an answer. Amelia now emulates that capability so that she can quickly and reliably retrieve information across a wider and more complex set of knowledge.
  • Emotional responsiveness:Repeatedly, research shows that a better customer experience is directly tied to empathy shown by the agent throughout the interaction. In addition to an EQ quotient, Amelia now has a mood and a personality vector in a 3-dimensional emotional space. This enables her to keep track of every client interaction and allows her to adapt her responses accordingly so that she delivers personal service to every customer.

In addition to progress in humanizing Amelia, architectural changes have been made to the platform in order to ensure that she is capable of supporting rapidly rising market demand. The entire backbone has been rewritten ensuring Amelia deployments can scale seamlessly and remain resilient throughout extreme peaks in volume.