Chatbots Do Not Need Turing

      No Comments on Chatbots Do Not Need Turing

 

Making technology personable and kind is a goal of many developers, and while making it personable would arguably make a user’s experience better, I don’t think it’s important enough to convince users that the thing you’re talking to is human. I’m mostly referring to what are called Chatbots. What is a chatbot? A chatbot, as defined by Wikipedia is a computer program that conducts conversation, hopefully making it natural enough to simulate human partner, and passing the Turing test, a test of a computer program to act intelligently enough to appear equal to a human. Chatbots are used across business, often for the purpose of customer service or collecting information from users, but are known mostly in popular culture because of entertainment programs like the now advertisement filled Cleverbot (it may have always been this way, but I remember in my younger age trying to annoy one of these robots on a blank page with just my queries and it’s responses).

The goal of these chatbots is to eventually be ‘intelligent’ or at least become convincing enough to talk with most people without a problem. The hunt for creating intelligent computers has been a long one, and one I don’t really care for. Chatbots that are meant for business operations are stand-ins for human beings for a process that one person could do themselves if there weren’t problems with confidentiality or business secrets. So really, these Chatbots don’t gain much from being able to respond intelligently, since those responses don’t really help the bot get any closer to completing the transaction it’s tasked with completing. Chatbots used for advertising a product don’t gain much from passing the Turing test either (although that could be fun to read, a chatlog between Chatbots of two competing items arguing over which is best).

In my estimation, the business roles that chatbots fill, being virtual assistants, advertisers on social media sites, or for entertainment purposes, the only variant that gains from being able to suitably pass the Turing test needs to be those designed for entertainment purposes. The problem with this is apparent to anyone who has interacted with one of these types of bots: most of the entertainment comes from the rigid, unnatural responses and from the weird responses from subjects out of left field because the bot interprets words in a way that the user didn’t intend. Making the robot seem like a broken mess is way more entertaining than the robot working properly.

Alan Turing’s Turing test for artificial intelligence doesn’t need to be passed for a Chatbot to be useful, and even if it was a robot that could pass the test, that doesn’t have any bearing on whether or not the bot can complete its task with greater efficiency for lower costs. Attempts to increase the thinking of a machine without practical application seems like an unneeded expenditure of resources. It’s probably best to leave complex and rationalizing thought to humans, no matter how rad a robot philosopher sounds.

 

Leave a Reply

Your email address will not be published. Required fields are marked *