* I will be continually adding onto this blog post periodically
What is a Chatbot?
- A chatbot is a service, powered by rules and sometimes artificial intelligence, that you interact with via a chat interface. (Schlicht, 2016)
- Automated computer program that simulates online conversations with people to answer questions or perform tasks. (Knowledge@Wharton, 2016)
- A chatbot system is a software program that interacts with users using natural language. Different terms have been used for a chatbot such as: machine conversation system, virtual agent, dialogue system, and chatterbot. The purpose of a chatbot system is to simulate a human conversation; the chatbot architecture integrates a language model and computational algorithms to emulate informal chat communication between a human user and a computer using natural language (Shawar & Atwell, 2007)
The opportunity for, and discussion around, Chatbots
- People are using messenger apps, where chatbots reside, more than they are using social networks. (BI Intelligence, 2016)
- Opportunity to reduce costs for companies. The chatbots envisioned by the tech industry combine artificial intelligence with voice recognition that relies on the way humans naturally speak. The goal is to create a situation where customers feel they are communicating with another human, rather than a piece of highly intelligent software, and in an environment that calls for little to no human operator intervention (Knowledge@Wharton, 2016). Indeed, research firm Gartner sees 33% of all customer service interactions as still needing a human intermediary by 2017,(Gartner.com, 2015) down from nearly 60% in 2014
- In general, the aim of chatbot designers should be: to build tools that
help people, facilitate their work, and their interaction with computers using natural
language; but not to replace the human role totally, or imitate human conversation perfectly (Shawar & Atwell, 2007)
- Bots are the beginning of micro apps on the backs of massive platforms which will lead to more focus and reach for startups and more delighted users (Batalion, 2015)
- Spend less time in development, E.g. by building a microapp on FB Messenger allows you to build one experience (and do away with writing for different mobile phone sizes, or OS) and access FB user rich profiles instantly
- Able to leverage richer UIs and multiple SaaS products to process input and focus more time on creating just enough of an experience to delight users
- Attracts users to stay within the platform eco-system where the chatbot resides – messaging is now the new platform. (Newman, 2016; Bayerque, 2016; Raziano, 2016)
- Increasing friction in getting consumers to download and use an app and quite costly as cost-per-install and paid acquisition marketing are increasing – the average global mobile user has = ~33 apps installed on his or her device and 12 apps used daily. 80% of the average global mobile user’s time is spent on 3 apps (Raziano, 2016)
Challenges surrounding chatbots
- Hype: There is so much hype around the potential of chatbots, that the general public’s expectations of what chatbots can do will exceed the reality of what they can actually do (Hobson, 2016). The reality also is that not every bot needs to be sophisticated, and it will depend on the objective / outcome that the chatbot is built for. If it is meant as an informational service, there is very little point in building a bot to conduct a conversation with a user, when the bot is meant to be transaction-based
- Developing a chatbot MVP (Minimum Viable Product): Software companies are used to developing a MVP, or a minimum standard version of the product that consumers would be willing to buy, to test in the market. However, for a good user experience, the product will likely need to have more accurate natural language processing and information before a MVP can be developed, which may mean that chatbots could require more capital than a traditional web or mobile app, where good frameworks are more commonly available (May, 2016)
What do consumers want in chatbots?
Based on a survey of 1,000 consumers in the UK in May 2016, the top perceived benefits of chatbots (MyClever Agency, 2016) was getting instant response and quick answers to simple questions. Interestingly, friendliness and approachability was not an important benefit for consumers, which indicates that customers want bots to enable efficient and accessible transactions.
Barriers to chatbot adoption by consumers
Supporting that view is consumers’ feedback on barriers to chatbot usage, and their biggest concern was that chatbots would not understand their questions. Chatbots that were incapable of friendly ‘chat’ was not a barrier at all for respondents.
Types of Chatbots
- Only responds to very specific commands. If you say the wrong thing, it doesn’t know what you mean. Only as smart as you program it to be (Schlicht, 2016)
- Understands language, not just commands. Gets continuously smarter as it learns from the conversations it has with people (Schlicht, 2016)
Who’s trying to attract/facilitate Chatbot developers?
- Google with Allo (Fulay and Adan, 2016), incorporating
- Smart Reply – respond to messages without typing a single word. Learns over time and will show suggestions that are in the user’s style
- Google Assistant – help to find information and complete tasks, e.g. book a dinner reservation, flight status
- Facebook with bots for Messenger (Marcus, 2016; Facebook Developers, 2016)
- Send / Receive API – ability to send and receive text, images, and rich bubbles with Call To Actions (CTAs)
- Generic Message Templates – structured messages with CTAs, horizontal scroll, urls, and postbacks.
- Natural Language Assistance – using the wit.ai Bot Engine to create conversational bots that can automatically chat with users
- Tools to enable discovery of bots – plugins for websites, usernames and Messenger Codes, prominent search surface in Messenger, ability for Facebook News Feed ads to enable the opening of threads on Messenger, a new customer matching feature will allow messages that are usually sent through SMS to be sent on Messenger
- Microsoft with Microsoft Bot Framework, a set of tools to help developers build artificial intelligence bots (Knowledge@Wharton, 2016), comprising (Foley, 2016):
- Bot Builder software development kit (hosted on GitHub) – for those interested in building bots using C# or Node.js
- Bot Connector – for registering, connecting, publishing and managing bots to text/SMS, Office 365 mail, Skype, Slack, Telegram, kik and more
- Bot Directory – directory of bots developed with the Bot Framework.
- Kore Bots Platform, that includes (Kore, 2016)
- A Natural Language Processing engine
- Enterprise Administration and Security
- Kore Bot Store – pre-loaded with more than 130+ ready-to-use enterprise and personal bots that perform thousands of tasks. Enterprises can also select, customize and create an approved collection of bots for their organization’s own private enterprise bot store.
History and development of Chatbots
- Eliza, the first chatterbot ever coded, was then invented in 1966 by Joseph Weizenbaum. Eliza, using only 200 lines of code, imitated the language of a therapist. He intended ELIZA to be a parody of human conversation, yet suddenly users were confiding their deepest thoughts in Eliza
- What was made clear from these early inventions was that humans have a desire to communicate with technology in the same manner that we communicate with each other, but we simply lacked the technological knowledge for it to become a reality at that time
- For more on the history of Chatbots:
- How The New, Improved Chatbots Rewrite 50 Years Of Bot History (Newman, 2016)
- A short history of chatbots and artificial intelligence (Bayerque, 2016)
- From ELIZA to ALICE (Wallace, n.d.)
Applications of Chatbots
- Weather (Schlicht, 2016)
- News (Schlicht, 2016)
- Scheduling (Schlicht, 2016)
- Friend (Schlicht, 2016)
- eLearning (Knowledge@Wharton, 2016)
- Employment recruiters (HR) (Knowledge@Wharton, 2016)
- Food orders and delivery (Boulton, 2016)
- Personal assistant (Fulay and Adan, 2016)
- Law assistance (Baker, 2016, Donotpay.co.uk, 2016)
Questions to ask when considering developing a Chatbot (Knowledge@Wharton, 2016; Jalali, 2016)
- How do they impact the customer experience/interaction?
- How do they fit with the positioning of the brand?
- How relevant is it to the business?
- Have we evaluated the resources that need to be allocated?
- What will the success metrics look like?
- User Acquisition
- How will users discover the bot?
- User Experience / Functionality
- How well do they work?
- What will the UX interface be like?
- How frequent will your bot updates be?
- How will you create a “Minimum Viable Onboarding (MVO)”, i.e. show users how to interact with the bot and understand what the bot is good at, to set expectations on how “human” the bot will be?
Tips for Building Chatbots (Schlicht, 2016)
- Decide what problem your chatbot is going to solve
- Choose which platform your bot will live on (Facebook, Slack, etc)
- Set up a server to run your bot from
- Choose which service you will use to build your bot
- wit.ai (bought by Facebook)
- howdy’s botkit (raised $1.5+ mil in funding)
- api.ai (raised $8.6+ mil in funding)
- Chatfuel (Ycombinator company)
- IBM’s Watson
- Dexter (owned by Betaworks)
Other areas of reading to beef up this post:
- Application of chatbots in HR and its implications
- Who else is facilitating chatbot development (Slack, WeChat etc)
- Comparison of the top 5 chatbot toolkits
- Deeper evaluation of chatbots
- Chatbot architectures and languages
- Loebner Prize
- Turing Test