How Much NLP Do Bots Need?
Natural language processing (NLP) is gaining prominence thanks to the recent popularity of bots. As technology evolves, sophisticated…
Natural language processing (NLP) is gaining prominence thanks to the recent popularity of bots. As technology evolves, sophisticated devices are becoming affordable and widespread. What once was the privilege of geeks and nerds has now become part of the mainstream.
Mainstream population is only focused on getting things done. Majority of people have no interest, nor time, to focus on mastering technology. They prefer the convenience of being able to get things done using natural language.
Enter bots. Current iteration of the bot technology finds them living inside various messaging channels. The communication on the channel is occurring by sending and receiving plain text messages. Bots can receive those messages from users typing them. Or, users can speak and the technology will transcribe the speech into text messages. From the bot’s perspective, it’s the same difference.
Naturally, NLP Matters
Users prepare their messages using natural language (for example, English language). When bots receive those messages, they must be able to process them. That’s the point where NLP becomes an important part of the equation. NLP technology is necessary in order for bots to process and understand the message.
But NLP is quite tricky. Even after five decades of focused research, quality NLP still eludes us. Should we wait some more and see if NLP will finally deliver? Or maybe now is the time to examine how much NLP our bots actually need?
Real Life Example
‘Bot’ is a simpler way of saying ‘robot’. The word ‘robot’ is of Czech origin, and it means ‘drudgery’, ‘hard work’ (for more detailed history of the word ‘robot’, read about Karel Čapek). In more contemporary terms, we could say that ‘robot’ means automation. Drudgery, hard work and non-amusing chores is something every human would like to avoid. The best way to avoid that drudgery is by automating it and giving it to robots (bots). No wonder bots are becoming so popular everywhere.
Compare mechanical/electronic bots with human servants. Suppose you need to take care of your garden, but you have no time nor skills. You hire a gardener. So this gardener is now doing the drudgery and the chores needed to keep the garden tidy.
Let’s now compare our experiences related to hiring a good gardener vs hiring a bad gardener. With the good, qualified gardener, we would just let him do his job. There would be no need to interact with him. The only interaction would be during the hiring process. That’s when we’d establish the compensation and the ground rules. Also, that would be the time when we establish our expectations with the gardener.
But if we hire a bad, unqualified gardener, the experience will be different. Because this gardener lacks skills, we’d end up having a lot of interactions with him. He’d be asking us all kinds of inane questions, because he wouldn’t be sure how to proceed each step of the way. In the end, we’d feel annoyance and frustration, and would most likely fire the poor gardener.
Same principles apply when dealing with bots. A good bot is a servant that does not need much interaction with us. He knows his job, and is doing it to perfection. There is little need for us to interfere and meddle into his affairs.
A bad bot will always be nagging us. He would need hand holding every step of the way. And that is super annoying.
Back To NLP
So now that we see the parallels between human workforce and bots, how much NLP do we need? When dealing with a good gardener, we need just bare bones NLP, because we’re not teaching him how to do the job. With bad gardener, we need all NLP we could muster, because we’ll be teaching him a lot of intricate stuff.
From this we see that it only makes sense to hire good, qualified bots. And we see that those good bots do not need a lot of interaction. Meaning, we do not have to sweat bullets around providing such bot with advanced NLP capabilities. A good bot is always a ‘set it and forget it’ bot. Otherwise, why bother? We could most likely do the job better ourselves. Meaning, we could fall back to self-serve. And bots really suck at self-serve.
Intrigued? Want to learn more about the bot revolution? Read more detailed explanations here:
The Age of Self-Serve is Coming to an End
Only No Ux Is Good UX
Stop Building Lame Bots!
Four Types Of Bots
Is There A Downside To Conversational Interfaces?
Are Bots just a Fad? Are GUIs really Superior?
How to Design a Bot Protocol
Breaking The Fourth Wall In Software
Bots Are The Anti-Apps
Screens Are For Consumption, Not For Interaction