Humans are increasingly talking to chat bots. From Alexa to Siri and Google assistant, we ask machines to take notes, play music or perform some other mundane or tedious task.
We talk to them on our own, or sometimes encounter them while looking for customer assistance. Companies are increasingly using chat bots to solve customer problems. Either way, these artificially intelligent chat bots are seeping into our everyday lives and will play a huge role.
Why are we interacting more with artificially intelligent chat bots? And, what can we expect of them in the future? Let’s find out!
Rise of Chat Bots
Being able to effectively communicate with artificial intelligence is a game changer. It is the most significant advancement in computer interfaces since the 80s when the Apple Macintosh introduced the point-and-click, a visual interface, to the world. Now, we’re able to be vocal and have a more two-way relationship with technology.
Chat bots don’t just respond single dimensionally anymore. They actively socialize with us like humans. They say, “hi”, and tell us jokes and want us to trust them. For this reason, they’re eager to be useful and help us.
Meaghan Anderson, the Vice-President of HubSpot, best illustrates this with a household example. She says that her daughter said her first word at 9 months, started walking at 13 months, and giving Alexa commands at 15 months! She thinks her daughter is growing up in a world where people can just tell the universe what they want, and they will get it. Such has been the impact of chat bots.
Reasons for rise of Chat Bots
- Humans Interacting like bots
- Chat bots are charming
- Chat bots are effective
- Makes business sense
Humans Interacting Like Bots
This is the irony. We’re automating customer service because humans respond like robots anyway. They read of scripts or go into soliloquies that no one listens to, which were hardly useful to customers. One IT support executive explained the problem. It is difficult for customer care executives to be patient and talk their customers through simple problems endlessly. A chat bot, on the other hand, has infinite patience and can better help customers consistently.
Chat Bots are Charming
Chat bots tend to be upbeat. They use emojis and gifs to liven conversations. They show genuine interest by asking questions. Plus, they can please you in a variety of ways – by playing a song or telling a joke, etc. Chat bots are effective in making you happy. They know how to trigger your emotions. Some worry that this is manipulation. But many find it very charming.
Chat Bots are Effective
If designed well, chat bots are impressively effective. They can connect with hundreds and thousands of users at a time, exchanging millions of messages and productively catering to each of them. They enable personalization, while scaling up customer service.
Makes Business Sense
Chat bots don’t have to hired or paid. Once you’ve designed an effective chat bot, you can rely on it to serve your customers with little additional investment. Hence, companies like Japan’s Yamato Transport, uses it to schedule courier deliveries and enable real-time tracking facilities. The famous pizza chain Domino’s uses a chat bot to take orders. Some worry these artificial intelligence technologies and chat bots may lead to mass unemployment. But, if we look to the past, we can find examples where automation led to an increase of business and opportunities for people. For instance, when ATMS were introduced, banks saved money, clerks were free more tasks, and more branches were opened to expand services.
The Case of Woebot
A couple of years ago, a clinical research psychologist (Alison Darcy) from Stanford, wanted to help anxious and depressed people receive therapy. She was a former computer programmer. Hence, she was able to think up Woebot, a chatbot that relied on text messaging to take Cognitive Behavioural Therapy (CBT) to a wider audience.
Technically, CBT comprises of simple techniques. But it requires regular interactions with a therapist. Most people can either not afford it, or don’t have the time for it. Some were too embarrassed to seek treatment. This is in the United States. In other parts of the world where even basic healthcare is a challenge, mental healthcare is even more inaccessible. This is why, Woebot was so important and necessary.
Darcy joined forces with a team of psychologists and a leader in artificial intelligence (Andrew Ng) to realize her vision. Sometime later, Woebot was online and free for all to use. Around 50,000 people (more people than a psychologist helps in their lifetime) downloaded it in the first week. Young men, married women, divorcees, all found Woebot useful.
One study found that 70% of young adults, who used Woebot, were less anxious within 2 weeks. Some of Woebot’s users said that the chatbot felt like a real person and was available round the clock. It even initiated conversation and always stayed upbeat. Woebot deals with millions of messages every week from a wide range of users. Since it’s not a human, people feel less judged in receiving the help that they require.
One of the reasons for the success of chat bots like Woebot is that they are openly artificial. They don’t pretend to be human. For a long time, programmers wanted to pass the Turing Test. They wanted to fool humans into thinking that they were talking to another human.
Google’s Duplex even managed to do this when it called a hair salon for an appointment. The artificially intelligent chat bot spoke exactly like a real person with “umms” and appropriate, timely responses. The hair salon was fooled, but people didn’t take it lightly. They value authenticity in these matters. Its why chat bot makers are open about being artificial. If you ask Alexa if she’s alive, she’ll probably say that she is, but not in the way that you are.
Another reason we’re moving into the post-Turing Test era is because the test is quite difficult for chat bots to pass. Conversations with humans are loaded with idioms, metaphors and implied meanings that chat bots have a tough time dealing with. A chat bot may not understand what you mean if you said “that was a piece of cake” (meaning easy). It might take you literally.
Challenges for Chat Bot Designers
- Creating the right persona
- Picking the right responses
- Making chat bots emotionally responsive and responsible
Creating the Right Persona
I’ve already said that chat bots tend to have a lively personality. But, designing the character is more than just being upbeat. For instance, people like to joke around with chat bots by asking if they fart and what not. Users want to be entertained in such circumstances and don’t expect serious answers. At other times, they may even request for a joke. But, when they expect a genuine response, a joke may be off putting. They don’t want sarcasm when asking a serious question. No one would ask a chat bot anything if they felt the bot would make fun of them for not knowing.
Picking the Right Responses
A major challenge for chat bots is to narrow in on the accurate response quickly. You can search for something on Google and won’t mind seeing a list of possible results. But for a text chat bot or a voice assistant, this is not an option. No user wants to see a long list of options on chat or patiently listen to all possible options for every question they ask. Hence, algorithms that power the chat bots must be extremely effective in finding right responses, which the designers must communicate well in a manner that reflects the bot’s persona.
Making Chat Bots Emotionally Responsive & Responsible
Sure, chat bots can appear to be emotional. But, how long can the bot keep up it’s emotionality quotient. There is the possibility of a user discovering that the bot is not programmed to be as emotional as it seems to be. Clearly, we users know that bots are not emotionally real. Nonetheless, we buy into their emotionality because it serves our purposes. The challenge for designers then is to keep the illusion alive from rupturing.
Chat Bots for Better or Worse
We may be heading to a future in which only the rich will have access to any services that involve humans. Most of the tiny interactions we have in our social lives like ordering food at the restaurant are already becoming automated. Consider ordering food at McDonald’s. Some of their outlets have a screen where you can place an order without any human interaction.
Although, such a world seems dystopian, some are of the opinion that it will make many quality services accessible to a larger audience. Everyone can access to a quality artificial intelligence tutor or medical advisor, for instance, without discrimination or worrying about footing a huge bill.
There is also the fear of artificial intelligence bots invading our privacy. To personalize services, bots collect a lot of data from us that run the risk of exploitation. The devices through which they operate also generally have easy access to our homes – the place that traditionally has been private and where we could easily block out the rest of the world. Now, we allow devices easy access to our private spaces making them easily accessible to the external world.
Finally, there is the risk of artificial intelligence bots knowing us better than we do ourselves. When that happens, they may use their knowledge of us to manipulate us and sway us for their purposes. The possibilities for disaster seem endless. Nonetheless, chat bots are still becoming popular and being used more frequently by humans.