DCSIMG

Artificial intelligence to make life simpler, even for dummies

the Hollywood vision of artificial intelligence features evil machines hell-bent on the destruction of the human race.

The reality, though, will be more benevolent, according to an internationally renowned computer scientist who is developing AI technology for more helpful ends.

Professor Lee McCluskey of Huddersfield University is an expert in the field of AI known as knowledge engineering for automated planning.

This involves developing machines to carry out tasks by making decisions based on reasoning. His current project focuses on efforts to manage transport systems in a more efficient way.

He told the Yorkshire Post: “Within the next 50 years we won’t be where the science fiction films like I, Robot and The Terminator say.

“What we aim for in artificial intelligence is not what people write in popular fiction.”

Instead, industry experts are working on projects to challenge and reduce the complexity of systems that have built up in the developed world.

Prof McCluskey said: “We are trying to make it simpler in the sense that we give machines goals and service levels that they have to achieve. They go away and do it rather than us having to go into their world and grapple with the complexity of systems.

“For example, with transport systems, there are so many ways to survey traffic, such as automatic number-plate recognition, close -circuit television, loops which calculate speed, lights, message signs and extra lanes.

“These systems are getting too complicated for people to actually control at a command level.”

The idea of simplifying the interaction between humans and machines will be on the agenda at next year’s international conference on automated planning and scheduling, which takes place in Brazil.

Prof McCluskey will be one of the two programme chairs, along with Brian Williams, who is professor of aeronautics and astronautics at the Massachusetts Institute of Technology.

They will have the power to decide which research papers will be presented at the gathering and which workshops, demonstrations and practical activities will take place.

The conference is regarded as the main forum for research in automated planning, a field regarded as essential for manufacturing, space systems, software engineering, robotics, education, and entertainment.

A good example of automated planning can be found in the Mars Rover exploration robot, said Prof McCluskey.

He added: “You cannot really have humans writing loads and loads for it to do operationally.

“It’s got to be able to plan and schedule its own day, partly because Nasa can’t contact it all the time.

“You can get algorithms to automatically generate plans and schedules.”

Meanwhile, Prof McCluskey has recently won funding for an AI project to improve the management of transport systems.

He is bringing together a group of academics, universities, technology manufacturers and suppliers together to study the application of autonomic principles.

He said: “Our bodies have autonomic systems. They are self-managing, self-monitoring and self-healing. Our bodily systems manage themselves and we don’t have to think about managing them.

“There are lots of examples of complex control systems that are too complex and too costly for people to maintain. Just maintaining the software is a nightmare.”

Asked the wider question of whether there is an inherent risk in teaching computers to think for themselves, he said: “Computers are never going to think for themselves in the way we do. We do need to make them smarter.

“If we assume that we will continue to rely on them in the future, and the demand will grow for more capable computing support – in household equipment, mobile appliances, cars, traffic management, internet, hospitals, planes, weapons and factories – then my opinion is we need to radically change the way we build computer systems. We already spend too much time interfacing with them.”

He added: “Imagine that we decide on a new environment policy and we want all traffic systems to adhere to this policy so that car pollution levels never exceed a certain level.

“What we should be able to do is to give the policy to the computer system and let it come up with ways that it could implement the policy and ultimately let a human choose which of these ways.”

Drive towards thinking cars

professor Lee McCluskey hopes to win funding for another project – this time to teach vehicles how to think for themselves.

In partnership with academics from Edinburgh University, the Yorkshire professor has applied for a research grant to look into “embodying autonomous vehicles with learning capabilities that will enable them to do deliberative, long term planning”.

In artificial intelligence, all objects have states. According to experts, the conjunction of all of objects’ states constitutes the world state.

 

Comments

 
 

Back to the top of the page

 

X scottish independence image

Keep up-to-date with all the latest Referendum news