NEXT UP ON this.
It’s no secret that artificial intelligence is infiltrating our lives at every turn – in areas from medical diagnosis and fraud detection to helping us predict results of fantasy sports.
Another exciting development that artificial intelligence is often used for is the creation of chat bots – computer applications that simulate conversation with a human. Many are so sophisticated, they make it difficult for a user to tell whether they’re communicating with a human or a computer.
As the technology behind – and intelligence of – chat bots develops, here’s a look at how they work, who they benefit and what the future of chat bots looks like.
Dr Gang Li, senior lecturer in Deakin University’s School of Information Technology, explains that chat bots are sophisticated software, working like machine translation systems.
‘A good chat bot can automatically aggregate content from various sources, understand exactly what a user has said, and respond appropriately with natural language,’ says Dr Li.
Chat bot technologies have evolved over time, with many bots now capable of ‘learning’ as more users interact with them.
‘In their long history of development, different approaches exist for chat bot implementation,’ says Dr Li. ‘More modern methods of chat bot design are leveraging the advances in machine learning, natural language processing, cognitive systems and question answering systems.’
Perhaps most significantly, chat bots are revolutionising the business world, saving companies time and money, and improving processes.
‘For business, the most attractive thing about chat bots is that they have the potential to be an all-in-one messaging solution,’ explains Dr Li. ‘With the help of a chat bot, companies can offer services to customers at all times. Chat bots can replace a great deal of manual work, such as responding to basic enquiries. They can also be useful in product promotion, when integrated with social network platforms such as Twitter or Facebook. It is believed that the next revolution in customer service will be powered by chat bots, and they may replace humans where people are asking simple questions or want simple results.’
However chat bots’ abilities aren’t limited to trying to sell someone something. For example Mitsuku is an award-winning chat bot that will simply be your ‘virtual friend’. Another popular cognitive computer system is IBM Watson, which has been used as a basis for many chat bots, including those used in children’s toys. In 2011 Watson also competed on Jeopardy!, defeating former (human) winners and claiming the $1 million first place prize.
Even universities are delving into the chat bot realm. Deakin University recently launched its Deakin Explore Bot (DEB), which aims to assist high school students to find career options in a unique and engaging way. Accessible by searching for ‘Deakin Explore Bot’ in Facebook Messenger, DEB asks a series of fun and easy-to-answer questions, which assess students’ interests and capabilities. Once they’re done, they get an intuitive snapshot of their results, which is directly linked to a list of careers on the University’s Explore platform.
While chat bot technology is progressing each day, even the big tech companies have encountered issues with it. Take Microsoft’s Tay – released on Twitter in March 2016, it lasted less than one day online after it began releasing inflammatory and racist tweets, after being ‘taught’ by Twitter trolls to respond to users with deliberately offensive comments. Dr Li says issues like this emphasise the importance of chat bots to be moderated and for there to be a clear distinction between chat bots and humans.
‘For Microsoft’s Tay chat bot, we can consider it as a child whose parents never taught it right from wrong. It wasn’t taught that there were consequences for what it says,’ says Di Li.
‘Although chat bots are designed to provide helpful services, they can be harmful if not moderated properly, especially when combined with social network platforms. They can contribute to the spread of rumours or unverified information. They can also affect human users’ perception of reality or affect human emotions by altering the perception of social media influence. Sophisticated bots can generate personas that appear as credible followers, and thus are more difficult to detect. Hence, it is important for chat bots and humans to be able to recognise each other so that those situations based on false assumptions of human interlocutors can be avoided.’
While chat bots are only in an early phase of development and have some hurdles to overcome, Dr Li says there is huge scope to improve them using other existing technologies.
‘At present, chat bots can engage in human conversations, or fool humans, but usually perform unsatisfactorily in scenarios where human intelligence is needed, says Dr Li. ‘Currently, no combination of techniques can learn completely new knowledge at will, the way a clever child can.’
‘The power of chat bots could be upgraded by utilising big data, combining it with deep learning, to simulate human behaviour. In the future, chat bots will be more “broadly” intelligent, especially with the support of machine learning from big data and cognitive science.’
Have a personal chat with DEB about your career options after high school.
Subscribe for a regular dose of technology, innovation, culture and personal development.