What Users Really Want From Chatbots
A narrative report based on qualitative research and usability studies conducted on Chatbots, within the graduate program in HCI at Bentley University.
People really want less needless chat, more relevant answers and more helpful actions, as quickly as possible.
That means: infinite inhuman patience, intelligence, speed and precision, coupled with simple human speech that is minimal in quantity, but is maximally useful.
They also want:
- One single search bar that is a unified source for all user questions
- No separate Chatbots for separate departments or groups within a company
- A face or icon to represent the Chatbot
- To relax and ask in lazy ways, and still have Chatbot understand the question and be nice back
- For Chatbot to preempt what the user will ask next
This is what Chatbots will have to be able to do in the future, if they are to be fully adopted, and if they are to make maximum money for their creators:
Examples of answers wanted:
- Factual: what is the support for the legal case that I am preparing?
- Hypothetical: what are my chances of a 14% profit within a given period of time, if I invest in this particular startup?
- Projections: how can my company optimize its operational expenditures?
Examples of actions wanted:
- Write my research paper for me, comparing the prospects of 3 alternative energy sources, highly weighing financial risk for particular mid-size companies
- Write my business plan
- Do my taxes now
- Teach me Russian
- Decisions tools: build me a visual interactive tool for planning my wedding finances and events
All this and far beyond can be done, and is in the works right now.
Once a Chatbot, supported by AI, is capable of answering more questions and perform more actions than is reasonable to fit visually into the information architecture of an app, we will reach the tipping point, where conversational interface will be more efficient than an app interface. And it is bound to happen.
Of course, some apps will never be overtaken by Chatbots — simply because they do not require a conversation to use them, and, in fact, are much easier to use without any conversation. Either because they are very simple, or because they already know the users well enough that they offer them only what they need, when they need it, and any conversation would only muddle the interaction.
That people get frustrated with Chatbots now is natural: Chatbots still cannot talk very well, they cannot really answer anything, and they barely can do anything useful at all. Does that mean that Chatbots are a lost cause? Not at all. I just think that right now is not their time yet: they are out a little too early, and the AI workware is simply not ready for them. We should still develop their conversational skills as fast as possible, to be ready for when the AI is.
It is just a matter of time… Just a tiny bit of time. Once the AI hardware is robust and self-sufficient enough to be our personal tech jinnie — we will finally need chatting Chatbots as an interface, and we will need them ASAP.
So, get ready: Chatbots are coming!
