Robots that think and feel like humans will live among us in 10 years, says inventor of the virtual nervous system
- Comment was made by Dr Mark Sagar, who is the CEO of Soul Machines
- Soul Machines has created a series of emotionally intelligent virtual people
- Powering the avatars is advanced software, a type of virtual nervous system
- Dr Sagar believes robotics will soon catch up to the on screen avatars
Robots that can think and feel like people could soon be living among us, according to one expert.
An AI engineer, who has invented a ‘virtual nervous system’, believes it is only a matter of time before robotics hardware catches up to his software.
He says responsive robots could be a common sight in businesses and homes around the world within the next ten years.
Robots that can think and feel like people could soon be living among us, according to one expert. An AI engineer, who has invented a ‘virtual nervous system’, believes it is only a matter of time before robotics hardware catches up to his software (stock)
The comments were made by Dr Mark Sagar who is the CEO of Soul Machines, an AI company in New Zealand that specialises in creating hyper-realistic 3D avatars.
The Auckland-based firm has used computer graphics to develop stunning renderings of virtual people, that can mimic natural facial movements like blinking and smiling.
But behind this attractive exterior is advanced software, a type of virtual nervous system, that can learn and even mimic human emotions.
And Dr Sagar believes it won’t be long before developments in robotics technology allow AI software to take on a more humanoid physical form, according to reports in CNBC
Speaking to website, he said: ‘We are creating realistic adult avatars serving as virtual assistants.
‘You can use them to plug into existing systems like IBM Watson or Cortana — putting a face on a chatbot.
‘We have been working on the deepest aspect of the technology, biologically-inspired cognitive architectures. Simplified models of the brain.
‘Robotics technology is not really at the level of control that’s required.
‘Robotics materials will have to get to the point where we can start creating realistic simulations.
‘The cost of doing that is really high.’
One application for this technology is Nadia, a virtual assistant voiced by Hollywood actress Cate Blanchett.
Nadia, launched in February this year, assists disabled people in accessing the New Zealand government’s National Disability Insurance Scheme.
It will take at least 12 months and a great deal of interactions with Nadia until she is fully operational, but she can already understand thousands of questions put to her and will answer with clear and simple responses.
And if robots can be created that can express the same range of subtle human-like responses and facial gestures, this could allow virtual assistants like Nadia to aide people face to face, rather than just from a computer screen.
Dr Sagar’s team at the University of Auckland hit headlines in 2014 for their work on BabyX.
One application for this technology is Nadia (pictured), launched in February this year, who assists disabled people in accessing the New Zealand government’s National Disability Insurance Scheme
The virtual assistant (left) is voiced by Hollywood actress Cate Blanchett (right), who donated her time to the project. It will take at least 12 months and a great deal of interactions with Nadia until she is fully operational, but she can already understand thousands of questions put to her
The virtual infant learns and acts just like a real baby, an important step to creating artificially intelligent brains.
Robots given these kinds of brains would be able to respond in the same way as the computer programmes.
Researchers programmed the brain to respond to certain commands, and to use recognition tools to allow it to identify words and images.
The artificial baby girl was developed by the university’s Bioengineering Institute Laboratory for Animate Technologies.
The virtual baby’s ‘brain’ is composed of algorithms that deduce what is good and bad.
BabyX (pictured), created at The University of Auckland in 2014, can recognise words and images. When spoken to the machine responds just like a real baby would. Using algorithms BabyX is trained to respond to certain situations
This enables BabyX to learn how to respond to interactions just like a real baby.
For example when a researcher holds up the word ‘milk’, the baby identifies the letters and says the word.
The researchers then praised the baby verbally, which releases virtual dopamine.
The baby learnt that correctly identifying words like ‘milk’ is good, and learns to do so more in future.
Reinforcement learning like this, similar to a real baby, helps BabyX decide how to react to certain situations.