What should a robot read to understand humans?
I speak to a robot every day…more than one,usually.
No, I don’t work at Westworld. 😉
This is how I define a robot at another blog of mine, The Measured Circle:
A robot is something created by humans (directly or indirectly) that performs tasks (autonomously or not) done by humans (or, more broadly, by other animals…a robot dog, for example, would perform work done by living dogs, including providing companionship).
The word may conjure up an image of a mechanical man, perhaps clunky and made of metal. The way we use the term at The Measured Circle, it would include software performing human tasks, and non-anthropomorphic devices like an answering machine or a calculator.
On the Robot Beat presents news about our creations that are, even in small ways, replacing us.
So, certainly, Alexa, whether on our
definitely qualifies. I talk to our Echo (family room) and Dot (bedroom) every day, and I bring our Tap to work, so five days a week on that. I rarely talk to our Fire TV or Fire TV Stick (we have both), but that may change when we can ask Alexa to open apps or show us shows.
I also talk to “OK, Google” on my Samsung S7 Edge.
They are all getting better…but I have to adjust my conversational style to fit what they understand best.
There’s a giant…well, let’s call it a “charms race” rather than an “arms race” to get digital assistants to have more natural conversation.
Progress is rapid…but there is a long way to go.
IBM’s Watson read all of Project Gutenberg (which prove to be confusing later on about what was fiction and what wasn’t, as I recall from reading a book on the project).
In this recent
Huffington Post article by Maddie Crum
it’s reported that Google used romance novels to try to get a bot to figure out how to put sentences together.
They liked that it was “formulaic”, and thought it was between the complex sentences of literature and the simplistic construction of children’s books.
It’s an interesting choice.
They aren’t really trying to get their bot to understand humans…just to be able to construct more natural seeming conversation.
They want it to see what type of response follows what, that sort of thing.
What would I have a robot read if I wanted it to make good conversation?
I think I’d go with Stephen King.
I’ve found that King writes people in a way that seems familiar to me, realistic to me.
You might have to pick and choose a bit…not every Stephen King book would work, or character.
You could approach this several ways.
You could have your bot memorize millions of books, and search for actual matches.
That’s going to produce some bizarre results from time to time. 🙂
Another way would be for it to figure out patterns…if the response to a question is “article, adjective, adjective, verb, noun”, that’s something a robot could learn. It would have to be able to either group words within that pattern by understanding their meanings to some extent, or just by grabbing groupings (a “gaze” might be “longing”, “intense”, or “steady”, for example).
My guess is that we’ll see rapid progress in the next two years and achieve bots passing the Turing test (basically, being convincingly human in conversation) within the next five.
I don’t think that means they need to think like us or feel like us. I think a way to fake that will be found by someone, and probably more than someone.
I would venture to say that most humans don’t think through everything they say, or even anywhere close to most of what they say. “How are you?” “Fine.” Much of it is rote, and more about social interaction than actual meaning.
That’s what bots will need to do, while still answering your question or performing the task you requested.
I don’t know how that’s going to happen, but I think it will.
Once they can handle conversation, they’ll be much better able to write fiction. They are getting there…not great fiction, but passable.
I found myself repeating the same stories over and over again. I use some of the same phrases repeatedly in this blog…sometimes by actual copy and paste (the introduction to my look ahead to the next month’s books, for example, tweaked each month. The statement at the bottom of each post is another).
Most of what I write here is new and spontaneous, but a blog like this is different from training a topic or answering a question.
I know some people recognize that I am telling the same story to make a point…and they are okay with that. I’m not trying to fool people…if it’s a good explanation, it’s a good explanation.
However, I don’t say it exactly the same way, mostly. I deliberately change a word or two, or word order, so that it doesn’t become monotonous.
This post, honestly, is just me musing. 🙂 I suppose, if I were a bot, it would have been more focused…and stiffer, most likely. 😉
What do you think? What would books would have a robot read to learn how to sound human?
Here are some more Goodreads deals:
Check the price before you click or tap that Buy button…may not apply to your country, and may have changed. I continue to be very impressed with these Goodreads deals, and I’m happy to be able to share them with you!
Join thousands of readers and try the free ILMK magazine at Flipboard!
All aboard our new The Measured Circle’s Geek Time Trip at The History Project!
*When you shop at AmazonSmile, half a percent of your purchase price on eligible items goes to a non-profit you choose. It will feel just like shopping at Amazon: you’ll be using your same account. The one thing for you that is different is that you pick a non-profit the first time you go (which you can change whenever you want)…and the good feeling you’ll get. Shop ’til you help! This post by Bufo Calvin originally appeared in the I Love My Kindle blog. To support this or other blogs/organizations, buy Amazon Gift Cards from a link on the site, then use those to buy your items. There will be no cost to you, and a benefit to them.