Experts predict that it won’t be long before software convincingly passes the Turing test. Before the end of the decade, it might not be possible to tell the difference between chatting to a bot and a real, live human being.
For many, such a future throws up dystopian images of a humanity increasingly cut off from itself. However, that’s not necessarily the way it has to go. In fact, chatting to robots could be one of the best things that ever happened from a number of perspectives.
Human-robot interfaces are growing across the planet. The investment community believes that companies that manufacture robots are likely to see wild gains in their value over the coming years, as prices fall, and capabilities rise.
However, the real gains will be cognitive. Machines will be able to do things with their “minds” that they never could previously.
Tell Them All Your Problems
Voice assistants are already showing us what this could look like. Even today, you can ask your smartphone virtually any question you like, and it’ll provide a quick answer for you, either verbal or by offering a link to a website.
However, in the future, they’re going to become far more communicative. Machines are building “common sense” which is giving them more context for the things that they say.
For example, you might want to talk to your machine friend about how you feel (instead of going to your romantic partner and stressing them out). To do this, you’d fire up an app and then either start talking or typing. This “therapy bot” would then respond to you in a way that’s appropriate for your experience.
The illusion that you’re talking to someone real breaks down quickly, though, under current systems. At present, bots make obvious and blatant mistakes that betray their electrical minds. But that needn’t always be the case. Bots are learning more common sense and putting their responses into context. Researchers are enabling them to develop working memories, giving them the ability to answer questions in a way that makes sense for the rest of the conversation.
Let’s say, for instance, that you’re talking to a bot about whether you should take a new job. You may say something like, “I’m worried that I won’t enjoy my new job. In my last job, I didn’t get along with people. What should I do?”
The bot might respond with something like “You can trial the job for a few months and see how it works out for you. Then, if you don’t like it, you can do something else.”
On the face of it, this response seems reasonable enough. However, if the person previously told the bot that they had anxiety about being out of work, then it doesn’t make a lot of sense. It would be better to say something like,
“Tell me more about the reasons why you didn’t get on with people in your last job.” This way, the bot can get to the bottom of what caused the previous issues and then make more tailored recommendations.
People Feel Freer Talking To Bots
Telling a robot all about your problems may help some people who are reluctant to go into therapy. There’s now considerable evidence that individuals with mental health issues prefer to tell computers how they feel compared to professionals.
In Japan, for instance, researchers wanted to see whether a realistic human look alike robot called ERICA could get people talking more. The majority of participants preferred to converse with a human therapist compared to a robot when talking about positive things, but only 30 percent enjoyed conventional interactions when discussing negative things.
Study participants, for instance, preferred talking to humanoid robots about jealousy or anger. They also liked talking about emotional pain, particularly with smaller, cuter robots.
The researchers aren’t sure why people prefer robots so much in therapy sessions, but they believe that it could have something to do with social hierarchy. Compared to regular counselling, people see robots as far less dominating. They are further down the scale of social strata.
Another hypothesis is that it is simply an alternative way for people to feel anonymous — a proven counselling technique which helps people to express their thoughts and emotions more freely. People don’t want to feel judged, and that’s precisely what a robot therapist gives them.
People Prefer Talking To Robots Over Their Managers
The benefits of robots may also be felt in the workplace. A recent study by Oracle found that 68 percent of employees would prefer to tell a machine how they feel, instead of going to their manager. Most don’t want to share stress- or anxiety-related issues with a superior for fear that it might impede their progress up the ranks. At the same time, though, they want help.
Oracle’s study involved more than 12,000 people, HR leaders and managers, as well as C-suite executives from a range of countries. It found that COVID-19 had increased the amount of stress people experience at work and this, in turn, had increased the number of people looking for mental health support as a result of their job. Around 78 percent of participants called 2020 their most stressful year ever, with more than 75 percent saying that artificial intelligence products helped them through it.
At present, the ability of AI to help people deal with their emotions is limited. There are apps that combine AI elements with other therapies, but they aren’t yet comprehensive. However, developers believe that they will be able to offer employees machine-based support by the end of the decade that offers the same benefits as conventional therapists.
Bots On Website Chat
We’re also seeing the rise of conversational software on business websites. Companies want to use bots to negate the need to maintain vast call centers to deal with customer queries (most of which are just the same things over and over again).
As click4assistance.co.uk points out, companies can carefully tailor these bots to get the most out of them. Instead of requiring them to engage in natural conversation with customers, these early chat selection bots simply get users to provide direct responses to the bot. The software then presents them with the next question to move them closer to resolution.
The concept sounds simple, but it is actually highly effective. If the bot can answer their query, it will. If it cannot, it will forward them to a customer service rep who can.
You can think of this concept as being similar to an old-fashioned call minder or receptionist. This individual would take calls and then direct them to the appropriate member of staff.
Robot chat bots do a similar thing. They take incoming calls and then redirect them to the person most likely to be able to answer the query.
How Much Better Can Conversational Software Get? Deloitte.com recently published an investigation into how conversational AI was improving and where they expect it to be by 2025. The market for such bots, they say, is growing. And the number of patents being filed that relate to conversational artificial intelligence is also increasing dramatically. But does it have much further to run?
Experts break down progress along five vectors:
- Training of conversational agents
- The ability of agents to handle complex conversations
- The degree of personalization to the customer query
- Improvements in voice assistance
- Improvements in virtual assistance ensembles.
Conversational agents, researchers suggest, require training on the content of the domain in which people use them. Call centre chatbots, for instance, would benefit from training on company FAQs and previous conversations with human reps.
Training currently takes up to a year, according to researchers. But improvements in learning methods could reduce the amount of data required. It might be possible to mimic human learning speed — perhaps a couple of weeks.
Deloitte’s research also sees chatbots getting better at handling complex conversations. They expect that they will gain more nuanced understandings of things like timetabling and calendar entries. For instance, a person might be able to give a command such as “please book me a taxi 90 minutes before my flight is due to take off.”
At the moment, a chatbot might look at a request like this with confusion. By 2025, it should have enough contextual knowledge to handle it.
We’re also likely to see increasing personalization. Chatbots, for instance, may be able to adapt the way they interact with customers based on the immediate needs of the business.
For example, if volumes are high, chatbots may respond by curtailing conversations or moving customers towards resolutions in a different manner. The same applies if a deadline is approaching and customers need to take action right now.
Chatbots may also adjust the pace at which they talk if they detect that they are speaking to an elderly person. Slowing down the conversation may improve comprehension and overall user satisfaction.
Takeaway
In summary, many people already prefer to talk to robots instead of human beings in various settings. As software capabilities increase, this process will likely accelerate.
