Don’t worry about the rise of the robots – the future will still be human
For many people, artificial intelligence (AI) and robotics can be slightly unsettling and, in the workplace, there can be concerns about the possible impact on jobs.
In reality, AI has the potential to enhance and improve our everyday lives and jobs. The type of research and applications I’m involved in with colleagues at Heriot-watt University, counterparts in industry, and peers across the world is designed to support human intelligence.
We believe that AI should be thought of as automating some of the repetitive, annoying tasks most of us face at times in our jobs. Technology can remove the drudgery and free up time to concentrate on more difficult or complex problems that require human intelligence, or give space to be more creative.
Automation has always happened; AI is just a different type. AI is not suddenly going to replace a large percentage of the workforce. I believe it’s going to support people to make certain tasks faster and easier. That’s why events such as the Centre for Work Based Learning’s symposium in November, #Ahumanfuture, are vital to help employers and employees understand how AI can work with them and for them, rather than be something to be wary of. The symposium recognises that the world of work is changing, partly due to AI, and new skills and knowledge are required.
There are many ways AI can support on-the-job training and I’m involved in work that can be adapted and used effectively by employers and employees. My focus is on conversation to aid human-robot interaction and the use of natural language. At Heriot Watt, we’re working on building conversational agents that people can talk to in everyday language and interact with for a variety of purposes, such as learning a range of new skills.
One of our projects in technology enhanced learning (TEL) involved the creation of virtual characters that people could speak to and interact with using natural language. One example is a language-teaching system where people can test their ability to do a certain task, such as going into a shop. Instead of interacting with a human tutor, the individual would talk to the graphical character, almost like a video game.
Such virtual characters can educate people in various ways, perhaps to train customer service employees in a contact centre. This type of simulated learning can happen anywhere and at any time – one of the big advantages of TEL.
Another TEL initiative has been designed to help autistic children by developing social interaction skills that could be transferred from the virtual to the real world. The ECHOES project created a virtual character exploring such issues as shared attention and eye gaze to encourage a group of children to play more and experiment with new things. The technology is now being worked on by the Institute of Education in London.
Recently, we’ve been involved in the Amazon Alexa challenge with our system, Alana, designed as an entertaining and engaging experience for chatting about a range of topics. Alana is an advanced form of AI software that understands and responds to human conversation, and her understanding of language is still developing. Its potential is being recognised as Alana has reached the final of this year’s international Amazon Alexa Prize, a $3.5 million university challenge to advance human-computer interaction.
Our Alana system can be used by everyone in North America who has an Amazon Echo. It’s easy to see how Alana can be expanded or modified for educational purposes. Through conversations with Alana I’ve discovered things I didn’t know and you may be pleasantly surprised by her knowledge. We’re hoping to spin-out a company using our Alana technology as an AI engine which could be used for many different applications, including TEL.
In the workplace, Alana could give conversational access to information relevant to certain organisations through a set of frequently asked questions that can be easily indexed. It could also be used to build task-related dialogues to allow certain aspects of workflow to be done using conversation. In particular, it could be good for automating very repetitive tasks, giving employees time to deal with more complex situations. People interested in Alana’s progress on Twitter can follow it at the @alanathebot account.
Overall, our research is building the kind of conversational experiences that would be ideal for workplace learning by allowing people to simulate particular problems and issues and build systems to respond.
Conversational AI can save time and resources by allowing training to happen at any location without having to set up physical meetings to bring people together. People can instead interact with AI through their laptops or phones which saves a lot of admin time, and they can train over and over again without others being involved.
To give an example of conversational AI being used by industry, Heriot-watt has a new project called ORCA – the Offshore Robotics Centre for Certification of Assets – which brings together five academic institutions and more than 30 industrial and innovation partners. Part of that project sees conversational systems interacting with robots that inspect oil rigs and pipelines, allowing people to better control the robots and understand more rapidly what the robot is sensing and seeing.
Skills 4.0: Work-based Learning For #Ahumanfuture is the theme for this year’s Centre for Work Based Learning in Glasgow on 6 November. Oliver Lemon is a professor in Heriotwatt University’s computer science department