Microsoft's approach to artificial intelligence (AI) is not trying to replace people with machines, but to empower humans to achieve more for themselves and for their businesses, according to Chris Bishop, scientist and lab director at Microsoft Cambridge.
Bishop said at the company's Future Decoded conference in London that his team follows three key three principles when building AI systems: to reinvent productivity and business processes; to build an intelligent cloud platform; and to create more personal computing.
Bishop also stressed the importance of laying down precedents. "AI is a very powerful and broadly applicable technology, and over the coming decade will transform nearly every aspect of our lives," he said.
"It's not surprising that there's a lot of debate and discussion about the future implications of AI in the long term for our society.
"When we create AI, we have a number of core design principles. AI should augment human abilities and experiences, AI should be trustworthy, and AI should be inclusive and respectful."
Bishop also mentioned Microsoft's "four pillars" in building AI systems: agents, applications, services and infrastructure.
A digital agent seeks to reimagine the customer service experience as a "frictionless" conversation between customer and agent that provides a seamless integration with a contextual dialogue, according to Bishop.
Current chatbots seek only to respond in a plausible way to each stage of a conversation, but Bishop claimed that a Microsoft chat agent does something "more challenging".
"The agent doesn't just say something plausible. It says something useful," he said.
Microsoft demonstrated software based on the firm's Kinect gaming motion controller that can segment kidneys for surgery, effectively telling surgeons where to cut.
"We're not replacing the radiologist or replacing the doctor. The capabilities of machines are complementary to the capabilities of people. We still leave the human expert to diagnose the tumour and to plan the treatment," he said.
Bishop then invited conference attendees onto the stage to translate freely and in real time between English, French, Chinese and (slightly embarrassingly for a 2016 IT conference) Klingon, again using an AI-based application.
Bishop explained that Microsoft's approach to AI is not to program a computer to be intelligent.
"That was tried in the 1950s, and didn't work very well. Instead, machine learning lets the computer learn from experience and is trained using large amounts of data," he said.
Microsoft claims to put a number of checks and balances in place as the company attempts to create AI systems responsibly, but as automation invades traditional job markets it remains to be seen how realistic the on-paper directives will feel when the reality of AI and machine learning starts to bite.
RISC OS 5 to form the basis of RISC OS Open after Castle Technology sells to RISC OS Developments
A smartphone maker fiddling its benchmarking scores? That's unusual, isn't it?
'We are making good progress on 10nm,' claims Intel
Engineer calculates that Chengdu's plan to replace streetlights with artificial moonlight would cost $100bn