IBM and the University of Michigan have announced they are developing a computer that can respond to conversation in the same way humans speak to each other.
Project Sapphire is a $4.5 million (3.14 million) undertaking that will first see computers act as an academic advisor for undergraduate computer science and engineering students at the university. The research into how smart machines interact with people will then be applied to more situations where setting and meeting goals are paramount.
The project will begin by collating human-to-human conversations between students and their advisors, which will then be used to train the computers in natural conversation and problem-solving.
"Natural conversations bring in so many different aspects of human intelligenceknowledge, context, goals and emotion, for instance. In many ways, to build a versatile conversational system is a grand challenge for artificial intelligence," said Satinder Singh Baveja, professor of computer science and engineering at the University of Michigan and director of its AI Lab.
The computers will learn using probability as the basis of reasoning, rather than context. This means things like knowledge, language understanding, emotion analysis and learning can be more accurate, IBM said.
"Human-to-machine interactions, similar to human-to-human conversations, are rarely confined to one question and one answer," said David Nahamoo, IBM fellow and chief technologist for conversational systems on the IBM Watson team. "They involve multiple turns of a conversation with responses that can be imprecise and unclear, making it difficult to simulate the human experience."
He explained that by partnering with the University of Michigan, IBM has the perfect opportunity to transform human-to-machine communication using AI technologies beyond the education sector.
"This collaboration marks the next chapter in a longstanding relationship between the University and IBM to place the power of cognitive technologies into the hands of the next generation of thinkers."
