Chatbots, computer programs that are designed to engage in online conversations with people, use artificial intelligence to mimic human speech patterns. While the concept may still be unfamiliar to some, they have in fact been with us for some time already. The Stanford NLP (natural language parser) has been around since 1990, so actually some of the technology is practically ancient. But chatbots are becoming increasingly used by businesses.
Now if you’ve shopped online or messaged a customer service representative about a minor issue, then the chances are you’ve been speaking with a chatbot. A number of major tech companies are now investing heavily in chatbots as they are offering substantial value due to their useful, interactive nature, giving what feels like a personal experience to the customer.
Poor customer service, whether it be down to an inarticulate complaint or a grumpy representative on the other end of the phone, has long been a niggling issue between businesses and their customers. Here, chatbots represent a potential solution that can have a significant positive effect on marketing and brand-building. Or so they say.
This is the dream we’re being sold. Brands however need to be careful that the sub-standard experience that most of these chatbots currently offer will far more likely damage their reputation than enhance it.
no umbrella needed, really? – Poncho’s weather chatbot
At the moment that is. As they improve and evolve, this will inevitably change for the better.
Here, we explore some of the many ways that chatbots are already being put to work, what positives and negatives they bring with them and what the future may hold.
Putting Chatbots to work
There is immense scope for the potential uses of chatbots across a wide range of industries, but for the time being it is mainly in the area of customer services where their impact is most being felt.
- Legal advice dispensed by the chatbot DoNotPay has helped drivers successfully overturn 160,000 unfair parking fines in London and New York in less than two years. The usually formulaic nature of the appeals process is well-suited to AI.
- In e-commerce, chatbots can be used for automated support and guidance in everything from providing product-specific details, to returning an unwanted pair of shoes or changing a delivery time.
- Facebook is broadening its offering to e-commerce providers, and now chatbots will allow businesses to interact with their customers efficiently and cost-effectively through the incredibly popular Messenger app. Naturally there is also an opportunity to build long-term relationships if customers make this their preferred communication channel.
- Banks are beginning to use chatbots to deal with queries that can range from lost or stolen credit cards to forgotten passwords.
- Then there is Assist, which aggregates a wide range of different apps so that if you want to send someone flowers, order a taxi, or book a hotel, you can do so depending on your preferences for timing or cost.
- Many prominent tech companies are developing Virtual Assistants, which are more evolved, intelligent programs. These include Apple’s Siri, Microsoft’s Cortana and OK Google. Siri made headlines recently when it was voice-activated to summon an ambulance by a woman who had dropped her handset.
- Amazon’s Echo, a device which some experts believe will be revolutionary in the field of ‘smart homes,’ has a built-in virtual assistant called Alexa that lets you shop (with Amazon of course), set alarms, create To-Do lists and even order a pizza from Dominos.
Benefits of using Chatbots
As far as businesses are concerned, the clearest benefit of employing chatbots, if employ is the right word, is the capacity for savings on staffing, and therefore expenditure. A touchy subject indeed.
If a chatbot is at the frontline, able to competently deal with relatively simple customer services requests, then a company can either cut down the number of staff it needs or move existing staff into different areas, such as fielding more complex customer service tasks instead. That’s because while a furniture chain’s chatbot can explain to a customer how to return an unwanted wardrobe, it can’t provide real-time advice on how to assemble it (yet).
Research to-date indicates that many consumers don’t really care if they speak to a person or a computer program, so long as their issue is resolved correctly and quickly. A chatbot can be programmed to speak in any language, can multitask, and once the initial development and implementation has been accounted for, its operating costs should be much lower than the salary of even a couple of employees.
Moreover, a chatbot can field repetitive questions about the same topic endlessly, without ever getting tired, irritable, or needing a toilet break. They are just as efficient at 3am as they are at 3pm. Companies measuring timings for responses and query resolution will find they are getting predictable results which are far faster than a human can deliver.
History is littered with scandals where a customer service representative has handled a situation badly, inadvertently triggered an argument with an aggrieved customer, who has then gone on to speak badly about the company across social media platforms. While there is still scope for reputation damage with a chatbot, the chances of situations such as this arising are greatly reduced.
Companies are adapting the bots by “enhancing” them with sentiment analysis software and putting the bots to use as a first line support.
As soon as the bot recognises any negative or strong sentiment, it can transfer the call to the human allowing for customer service to still employ empathy but only when needed.
Chatbots enable simplified communications between brands and their customers, and if used in conjunction with an app such as Facebook’s Messenger service, can form one long conversation that can be retained indefinitely. Within that communication, over time, there is the opportunity to harvest substantial valuable data which can help the brand meet the customer’s needs more effectively.
In future, we will see chatbots storing information about customers so that they can serve up products suited to their preferences, refer to them by nicknames or potentially, with diary permissions enabled, reminding them about family birthdays, anniversaries and other events.
Potential clouds on the horizon
Chatbots (at the time of writing this) are not as advanced as they could be and carry a significant risk of annoying anyone communicating with them.
A few months ago, Microsoft launched an experimental chatbot it called ‘Tay’ on Twitter. While Tay coped with the volume of communications well-enough, sadly it struggled with the content. Artificial intelligence, learning as it goes from human responses, can be trolled and Tay was rapidly taken offline after it began to make offensive remarks. Chatbots need to be taught what is appropriate communication and what is not – the nuances of human interaction. Yet in theory this instruction would not need to be ongoing or could at least be scaled down after a while.
Chatbots are not appropriate for every business, and as there can be significant costs involved in setting them up, a company must assess whether they are actually necessary first. Chatbots are best-suited to those businesses which engage in a lot of regular, repetitive communications with their customers, such as e-commerce stores or take-away services that have standardised menus.
As chatbots begin to take over more and more functions, there is a possibility they will one day remove the need for humans altogether in customer service roles. Assuming that does eventually prove to be the case, in a country such as the UK, where a very high proportion of people work in the service industries – one estimate has it that almost 400,000 jobs might be affected – this will need to be well-managed to ensure that those employees can move into more skilled and rewarding positions.
Out of control – Microsoft’s Tay shortly before it was taken down
“I’m sorry Dave, I’m afraid I can’t do that”
Perhaps the most pressing issue of all around chatbots is that of privacy. The more information people share with them, the more helpful they can be. But at what point might that level of awareness become invasive? Data security concerns will naturally arise. This is a grey area, and one where technology companies will need to listen to their customers in order to find the right balance.
The Chatbot Revolution
Predicting the future of technology is always a dangerous game, but some reasonable guesses can be made about what the next few years might hold for chatbots.
The immediate future will involve increasing numbers of businesses looking into human/AI hybrids (no Terminators hopefully), where a real person is at hand to deal with more complex customer service queries, or intervene should a conversation go off the rails. Individual businesses must decide for themselves the appropriate balance needed.
Provided that growth in chatbot usage is handled well, then overall employment levels in the UK’s extensive services industries may not be too badly affected, as redundant staff can move into other roles such as supervising chatbot interactions. We can also expect to see chatbots being increasingly introduced to a raft of other industries, from manufacturing and education to medicine and finance, as brands begin to develop trust in the technology.
Businesses that can successfully make chatbots work for them will potentially see rapid increases in their profitability, due to gains in time and productivity, while rates of customer satisfaction should also receive a boost.
Assuming that progress in chatbot technology continues to improve and business uptake grows alongside it, then within the next few years we are likely to see chatbots in use all over the web – or rather, we may not see them.
Ultimately, success will mean chatbots passing the Turing test – where it is impossible to differentiate in real-time, online conversations, between the computer program and an actual human. This will involve the chatbot having not just the ability to respond correctly but also to comprehend. It will also raise even more red flags about privacy, security and trust issues for web users.
However if all that hasn’t scared you off and you are looking to develop a chatbot or think that your business could benefit from automated customer service, then IBM Watson with its free API could be a very good place to start.