AI will rule but robots must get smarter to co-exist with humans
The Cambridge technology cluster’s growing reputation in Artificial Intelligence will make it easier to pull in much needed talent to the UK’s most innovative region according to Cambridge Consultants’ head of AI, Monty Barlow.
“Multinationals are looking at where to invest and this area is very high up in AI research and startups, with companies like PROWLER.io, said Barlow, who also heads up Digital Greenhouse at the innovation company.
Speaking to Business Weekly, Barlow said: “All the big players such as Amazon and Microsoft are setting up research labs here. It’s going to touch on everything.”
The product design innovator’s most recent disruptive technology programme – Deep Ray – started with the question: “how can we look through walls?”
Due to be launched next month in Munich, Deep Ray utilises deep learning where a machine learns as a human, to create a perfect image from pictures viewed through distorted glass.
This is one of the first developments from a partnership between Cambridge Consultants and California-based companies NetApp, which does data management, and NVIDIA, which specialises in GPU (graphics processing unit) computing.
“As humans we are really good at seeing through distorted glass, smoke or fog and filling in the gaps,” explains Sally Epstein, Machine Learning engineer at Cambridge Consultants. “In developing Deep Ray we wanted to see if we could get technology to do the same and push it one bit further to see if it could do better than humans.”
The technology is currently running on a laptop and is close to being embedded in a chip. Uses they see for it include to help the vision of a firefighter walking into a burning building or for sharpening CCTV images.
The company is also looking at applying it to other human senses, with applications including ultrasound. In setting up the partnership the companies believe that AI is on the verge of touching “each and every” industry just as electricity has over the past 100 years.
Barlow adds: “AI is currently at two stages: untrusted, where robots aren’t allowed to do anything other than follow strict rules and sophisticated, where it doesn’t matter if they get it wrong.
“There’s a corner where we all want to get to that we call ‘Advanced AI with impact’. If you can learn to feel you can trust the machine there are much better outcomes for society.“
Deep Ray follows the launch of a number of AI developments last month by the partnership, including Aficionado, which beats hand coded software hands down on a test to recognise types of music not heard before including Baroque, classical, jazz and ragtime.
The collaboration has also developed Vincent, which can complete an artwork that has been started with a human sketch. The system was trained to do this via 8000 examples of masterpieces from the last six centuries.
As well as art, the team sees this as having applications for the future of computer aided design and non creative uses such as with test cases for autonomous cars.
Deep learning has also been used to train an AI system to play Pac-Man without giving it the rules and seeing the choices it makes.
The aim is to train it to get towards making life and death decisions such as with search and rescue. NetApp’s role in the partnership is to power Cambridge Consultants’ AI capabilities, providing a platform that can handle the vast amounts of data required for predictive analytics, image-recognition and advanced searches.
NetApp works with customers to innovate in the cloud, including helping them move data and synchronise data between the cloud and a company’s own data centre. “We are not a typical AI company, but we activate and simplify the AI journey,” said Christian Lorentz, NetApp‘s senior product manager EMEA.
”Our data fabric is the glue to connecting all the solutions. This can can seamlessly, easily and safely move data where it’s needed.
“We can help companies modernise their infrastructure giving them more performance through flash technology.”
NVIDIA’s role is to “create a neural network and then pass large amounts of data through that to create the machine equivalent of the human experience,“ said Carlo Ruiz, the company’s Business Manager, DGX Platform for Deep Learning EMEA. “It’s truly a big bang what is happening with AI and it’s moving at a very fast pace.
“Ai has had a dramatic impact on healthcare and has also changed things like transportation. It’s being fuelled by deep learning, which we can compare loosely with the way the human brain works.”
This required eight of its highest performing processors, the equivalent of 140 conventional servers. While many fear AI will take jobs Barlow believes the opposite will happen.
“It will help people be safer and more productive in what they do. It will change the type of jobs people do but one of the things we need to do is go on using what’s best about human beings – the creativity and insight – and build tools that let machines and humans collaborate better.
“If a robot is going to co-exist with humans it’s got to be smart and predict what humans around it will do based upon its own actions.”
• PHOTOGRAPH SHOWS: Monty Barlow