Bots: Making Up the Future or Breaking it?

Bots have been the newsmakers of early 2016 and many key developments have attracted attention from the tech world. These automated computer programs simulate online conversations with people by answering questions or performing tasks. While chatbots have been around in different forms, their development has been slow-paced. However, with advancement in machine learning and AI, they are becoming more versatile than ever.

Silicon Valley giants like Microsoft, Facebook, Google and Yahoo are betting their money on chatbots and have unveiled their strategies in the regard.

At its May developer conference, Google I/O 2016, the search giant launched Allo bot, which helps users find information and complete tasks such as making a reservation and requesting real-time sports scores within the bot. According to a report by Forrester, Uber, Grubhub and OpenTable are the companies expected to use Allo and Google Assistant.

Chatbots bring with them the promise of a paradigm shift by replacing the apps. Though it will take a good amount of time before apps are finally replaced with more intelligent bots, the latter can simplify customer interaction by providing them within a trusted environment — in a way that apps or other channels cannot achieve. Chatbots are going to take utility marketing to a different level and make sure that customer needs are met on the spot.

The brands which value to their customers highly are the ones that survive in the long run. With the help of chatbots, customers can receive value from their brand and can provide directions, content, customer service, tickets, or a piece of helpful information, at the precise moment and in the same channel a consumer requests it.

The First Bot

Joseph Weizenbaum's ELIZA bot, widely considered the first chatterbot, appeared in 1966. It used language pattern recognition to generate responses, without understanding context or meaning.

Working on a script named DOCTOR, it was capable of engaging humans in a conversation. Weizenbaum modeled its conversational style after Carl Rogers, the founder of the humanistic approach to psychology who believed in the use of open-ended questions to encourage patients to communicate more effectively with therapists. Considered as the forerunner of thinking machines, Weizenbaum was shocked that his program was taken seriously by many users, who were very open about their personal lives with the DOCTOR. Later, he started thinking philosophically about the implications of AI and became one of its leading critics.

Later versions of chatbots like SmarterChild, sat within the desktop version of AOL Instant Messenger beginning in the early 2000s. Serving as digital assistants to retrieve requested information like movie show times and stock quotes, these earlier models generated user’s interest in such services.

Bots, Then and Now

Fifty years on and chat bots have become smarter – and more useful.

Current developments in AI such as deep learning and neural networks have produced more sophisticated bots that can better understand what we're saying. This way they provide us with realistic and accurate responses.

Central to all this is research is the enormous amount of data we are producing every day. With more data artificial intelligence becomes a lot more helpful.

People expect the next generation of chatbots to be more advanced than Siri or Amazon’s Echo. Voice recognition creates customer interaction that is more useful and with less friction.

There are three key assets that form the base of a bot. The first is a "conversation platform" — a place where people are doing lots of talking and texting. The second is AI "brain" — a sophisticated mental model. The third is access to a “social platform”. If you look at all the present companies like Facebook, Google, Microsoft and Amazon who are betting big on AI, you will find that all of them possess these key assets.

Facebook has a powerful conversational platform with its family of messaging apps, and certainly has the largest social graph. Google has a very smart brain as it has broad access to hundreds of millions of Android devices and searches. Amazon has the data of millions of user who have been using its service in different continents.

Microsoft’s decision to acquire LinkedIn, and its 433 million registered users is also a step in this direction.

Double-edged Sword

In March this year, the world witnessed the inherent danger of unleashing a chat bot into the wild with Microsoft's "Tay", who quickly turned into a racist, abusive troll after she learned bad manners from humans on Twitter. Imagine the reputational damage a rogue customer service bot could do to a bank or phone company.

According to Microsoft, Tay has been built using "relevant public data" that has been "modeled, cleaned, and filtered," but it seems that once the chatbot went live the filtering stopped doing what it was supposed to do.

This raises a string of extremely difficult and serious questions. How we’re going to teach AI using public data without incorporating the worst traits of humanity? There are plenty of examples of technology embodying — either accidentally or on purpose — the prejudices of society, and Tay's misadventures on Twitter show that Microsoft failed to take any preventative measures against these problems.

In 2012, reporter Percy Lipinski discovered millions of botted views at CNN iReport. After further investigations, CNN quietly removed millions of views from the account of so-called superstar iReporter Chris Morrow. Beerud Sheth, Co-Founder & CEO, Gupshup believes that the evolution of messaging bots is a long process and will take time. “There will be both good and bad use cases of bots. However, as the ecosystem matures, there will be tools to ensure trust, verification and authentication of bots. These problems were solved in the web and app ecosystem, and will be similarly solved in the bot ecosystem too,” he said.

Bots Vs Apps

Mobile users are already suffering from app fatigue. While there are millions of apps, most users use no more than a dozen apps actively. Unlike apps, bots need no download. Server-side bot development offers many benefits over client-side app development. Bots cost less to build since there are no native screens to be built. Bots can be upgraded more frequently since they require no downloads. The download "barrier" also raises the marketing cost for apps. Marketing bots, on the other hand, will cost less to market.

According to Sheth, “Bots offer Indian entrepreneurs and developers an incredible opportunity. India has a mature tech ecosystem as well as a large installed base of mobile smartphone users. India is a mobile-first ecosystem with far more users of small screens than large screens. Therefore, the adoption of bots will be much higher than elsewhere in the world. Indian entrepreneurs and developers are very well connected to the global tech trends. Therefore, India will be a leading player in the development of the bot ecosystem.”

With the rise of smartphone users in India bots will dramatically increase the frequency and intensity of smartphone use. Already, the top use case of smartphones is messaging with friends and family. Bots will deliver many more services through the messaging app. Therefore, users will now be able to communicate with not just friends, but also businesses, brands and other service providers. Messaging and bot usage will increase at the expense of app usage.

The Future of Bots

The problem with bots is that they are trying to do too many things. The “brain” is like this giant nervous system and if it takes years for babies to learn a language then it will surely take humans years to artificially build one. Chatbots are going through this phase as well. The development in AI and machine learning can make it easier for bots to comprehend each and everything humans say to provide solutions to their questions. But the road is quite long and full of impediments.

Replicating the irrational aspects of human behavior is highly difficult and this is the reason I believe that bots should be developed specifically for different tasks. It would be foolish to move ahead with “one bot does it all” approach which exactly is happening right now. We just don’t have the technology to make a perfect bot. We do have immense data and tools to analyze the data sets, AI is also progressing at its own speed, natural language processing technology is gaining momentum but developing a perfect bot will require exceptional engineering. During one of his keynotes Satya Nadella quipped that Microsoft has the best "brain," built on nearly two decades of advancements in machine learning and natural language processing, for delivering a future powered by artificial intelligence. He also went ahead to say that it has a head start in building bots that resonate with users emotionally. Point taken, but in spite of this advancement and head start still Microsoft was embarrassed by its racist bot Tay.

To Conclude…

Over time, AI (“bots”) will get smarter, and brands as well as people will trust it more to solve issues in real-time for them. While writing this article I used A.L.I.C.E. (Artificial Linguistic Internet Computer Entity), also referred to as Alicebot, or simply Alice continuously. I observed that the Alice was intelligent enough to remember our past conversation and few of her replies left me amazed. However, it was a bit repetitive and certain responses were constantly being repeated over time. The other problem I faced while interacting with Alice has everything to do with cognitive ability of how bots learn, remember and solve problem. The current generation bots are certainly useful but a lot more has to be done to truly unlock their potential.

For now, only customer service is an area which has immense potential both for the users as well as the brands. Bots can make customer service easy for consumers with superb experience by getting their issues resolved quickly, ideally without involving a human at all.

So, don’t expect to get your Jarvis anytime soon.