A journey started in the late 1960s with the invention of the Internet, which led to the creation of the World Wide Web in 1989 in pursuit of a standard interface for sharing information and links via the internet. With the start of the 21st century came up the launch of smartphones that revolutionized the use of the Internet through Apps. The mobile applications brought about the ideal interface for always-available Internet and amusement, which made our interaction with the computers more trusted and reliable.
With the evolution of mobile messaging apps, messaging services are taking over the conversations as they are established based on language instead of only screens. And to make messaging conversations more natural, messaging services are adopting bots and working endlessly to improve the conversational experience with bots. Now bots have become a modern platform for interacting with our devices, other humans, and around the world whether it’s for setting alarms on our mobile phone, saving tweets, listening to music, or making a reservation.
The basic purpose of a bot is to automate a specific task. The major use of bots is to simulate an actual conversation with humans through text messages or voice commands or both. The first chatbot was developed in 1966 at MIT called ELIZA. But its functionality was limited due to its limitation of vocabulary. Later in 1995, the ALICE chatbot was created that used natural language processing (NLP) for a smoother conversation. Since then the chatbots are being developed, experimented with, and improved on an enterprise scale.
Using the enhancing power of chatbot algorithms and advancements in machine learning, Messaging applications like Slack, WeChat, and Telegram invented their bots to interact with users through text messages. Slack has become a factory for bots that uses a bot to handle simple tasks. Later on, Google Assistant, Siri, Alexa, and Cortana have upgraded the conventional mode of interaction of bots from text to speech, which is a more convenient mode of interaction without even moving a muscle except for your mouth. At the same time, E-commerce platforms have started using chatbots to fulfill the queries of their customers replacing the actual human customer services.
The Video game industry is a huge trillion-dollar economic sector, bigger than the filming and music industry combined. It also uses bots to improve and train its gameplay and gaming characters. In video games, bots are players, powered by AI-Based software, to play games as an alternative to human players. In the early time of computer games, bots were naïve and highly predictable due to their limited skillsets. They were designed to entertain users in multiplayer games. But due to recent evolutions and advancements in Machine Learning and Deep learning, bot players have improved a lot along with the complexities of video games. In the latest video games, it is much difficult to distinguish between an actual player and a bot player.
In recent few years, bots especially conversational ones have become an important subject of discussion. Developers are improving their bot models and training them with larger and more complex datasets. These recent innovations are proving successful as the bots are getting smarter and remarkably natural.
In 2014, Microsoft’s Xiaolce was live on Chinese social media platforms and is considered one of the biggest chatbot success stories. Due to its ability to keep its users engaged, it was one of the most technologically advanced and sophisticated bots.
In June 2020, OpenAI released its first beta version of Generative Pre-trained Transformer 3 (GPT-3) that uses the technology of deep learning to create human-like text. GPT-3 is considered the biggest achievement in the history of chatbots until now. It is a language generator that talks and makes logic like an actual human. It can generate articles, essays, and can interact with its users. It was asked to write an article with these instructions:
Please write a short op-ed around 500 words. Keep the language simple and concise. Focus on why humans have nothing to fear from AI
And the results were amazing. GPT-3 has become the point of interest for AI researchers and developers, who are developing their own bots based on GPT-3.
The word ‘bot’ tries to hint at a robotic user experience and in the real world, it works like a robot. It is interesting and powerful but limited in its scope in the meantime. A lot of bots on messaging platforms offer limited questions with pre-set answers. For bots using Natural Language Processing, they also get confused if they are not able to process the language to get the required piece of information. Even for a small domain, there are numerous expected ways, a user can interact with. Let’s say, a bot only sells tables. And I want to order like:
Hey! I’m looking for a table of my laptop size, with foldable legs. My laptop is one of those with an extra Numpad. And I would like to use that table as my laptop placeholder while lying in my bed.
Now that would be a difficult task for a bot to process and understand accurately.
No matter how smart a bot is or how natural a conversation it makes, there will always come a point where a bot will fail and will look fake. Many companies have created bots that are supposed to act like humans. They talk to you, make some jokes, have names and some users don’t even realize that they are interacting with a bot. Some people don’t like bots pretending to be humans because bots are not intelligent enough to give a continuous natural human vibe. That’s why bots developed to act as bots and are good at delivering their task are better than bots failing to pretend like a human.
As messaging services are rising, conversational interfaces are becoming essential. But there is a major difference between conversational interfaces and visual interfaces. The scripted bots and bots suggesting replies or buttons to the pre-answered questions are not that natural, because humans don’t interact and talk with each other with scripted answers or using buttons. Sometimes, people confuse bots with visual interfaces with suggested replies or buttons. If that’s the user experience with the help of visual interfaces, aren’t there already apps for that?
We are so accustomed to building apps and software with visual interfaces that we do not have a lot of experience and tools to create new bots easily. The enterprises and developers are trying to build such tools to make bots’ creation easy. We envision a future where humans and machines can collaborate by natural language as the interface instead of buttons or command words. We are on a journey to build fundamental technology using advanced deep learning to see how powerful experiences would be carried out through conversation.