13 Facts About Chatbots You May Not Have Known

Everyone loves a good tidbit of trivia. Here at ZyraTalk, we're especially interested in fun facts about chatbots and artificial intelligence. Without further ado, let's dive into some obscure, well-known, and just plain wacky facts about chatbots.

13 facts about chatbots that you probably didn't know

1. The first chatbot was invented in 1966.

Joseph Weizenbaum created ELIZA, a computer program that simulated a human therapist, in 1966. An MIT professor and one of the computer scientists who led groundbreaking research on AI alongside the likes of Alan Turing and John McCarthy, Weizenbaum wanted to create a program that "tricked" people into thinking they were speaking with a human.

Much a like a modern chatbot, ELIZA operated using keywords that triggered a pre-programmed response. "Her" ability to ask open-ended questions through "DOCTOR" scripts was instrumental in making people believe they were chatting with a human.

2. Accomplished computer scientist Michael Mauldin coined the term "chatterbot" in 1994.

Even though chatbots had existed in some form for nearly 30 years prior, they wouldn't receive a proper name until the mid-90s, when Michael Mauldin invented Verbot. A portmanteau of "verbal" and "robot", Verbot evolved from Mauldin's first chatbot prototype, Julia.

Mauldin has since retired and now spends his time competing in the Robot Fighting League.

3. In 1972, psychiatrist and computer scientist Kenneth Colby created a chatbot that simulated paranoid schizophrenia.

When it took the Turing Test, this unique program succeeded in fooling human psychiatrists just over half the time. PARRY went head-to-head with ELIZA in the fall of 1972.

Their conversation was entertaining, to say the least, and ended with PARRY calling ELIZA "a real nag". You can still read a shortened version of the transcript.

While PARRY was undoubtedly iconic, it attracted criticism from other leaders in the field. Colby defended responded to some of the most common accusations in a 1974 memo. Here are a few of them:

• PARRY's natural language processing parameters were insufficient
• The program's portrayal of paranoia didn't model the underlying processes that trigger paranoia in people with schizophrenia
• PARRY did not divulge the causes of its paranoia

4. Invented in 1995, ALICE was one of the first chatbots as we know them today.

Fast forward over 2 decades and you'll find that ALICE isn't all that different from the chatbots we interact with today. ALICE uses natural language processing to provide credible responses. She and her creator Richard Wallace have won several awards for their ingenuity.

Ask ALICE a vague, open-ended question and she'll deflect like an expert. In other words, she'll answer the question with a question.

To borrow an example from the Ubisend blog, if you ask ALICE, "Why are apples red?" she might respond with, "Does everything need an explanation?"

Why not take her for a spin? You can still "talk" to ALICE at alice.pandorabots.com.

5. ALICE partly inspired the film Her, in which a man falls in love with his automated virtual assistant.

That effectively sums up the 2013 blockbuster Her, starring household names like Joaquin Phoenix and Scarlett Johannsen.

Johanssen voices Samantha, the automated digital assistant of Theodore, the movie's protagonist played by Joaquin Phoenix. To make a long story short and spoiler-free, Theo falls in love with Samantha. But it seems monogamy isn't part of Samantha's programming — she reveals to Theo that she's talking to over 8,000 other users and she's in love with over 600 of them.

Here's another fun fact: Samantha is named for the text-to-speech feature on early Mac computers.

6. It's possible to "break" a chatbot.

"Breaking" a chatbot doesn't make it malfunction. It simply outs it as an artificially intelligent entity.

If you're not sure whether you're talking to a human or a bot, you can ask a series of open-ended questions, like "How are you feeling?" to find out.

You'll usually receive a stock answer that will indicate you're speaking with a bot. Some bots are harder to break than others — and it turns out ALICE is one of them, even after all these years.

7. Experts predict that users won't be able to tell the difference between a human agent and a chatbot by the end of the decade.

In a 2016 interview with The Verge, Google's Ray Kurzweil predicted chatbots will be indistinguishable from humans by 2029. Certainly, chatbots have already come a long way since then. But only time will tell whether Kurzweil's prediction will come true.

At the time, Kurzweil and Google had teamed up to create a chatbot called Danielle, inspired by one of Kurzweil's original novels. No news yet on how that development is coming along.

Some of Kurzweil's other predictions are pretty bold. One day, says Kurzweil, programmable nano-devices will replace our cells. It's safe to say we're a long way from that, but it's not impossible. Not that long ago, devices like pacemakers seemed unachievable, too.

8. Soon, chatbots will be able to engage in conversation about virtually anything, including celebrity gossip.

If you've ever asked Siri to tell you a joke, you're probably not too surprised by this. But chatbots aren't exactly skilled at small talk.

None other than Amazon is looking to change that. In 2018, Amazon launched a competition called the Alexa Prize Socialbot Grand Challenge. The competition invites student teams to create a chatbot capable of gossiping and small talk.

Before long, you might not even need Twitter or Instagram to keep up with the latest celebrity scandals.

9. Q is a new automated assistant that's gender-neutral.

If you've ever used an automated personal assistant, you'll know you can program them to use traditionally male or female voices. Now the world has its very first gender-neutral assistant.

Audio researchers gathered voice data from non-binary people to create Q, the genderless voice. This is particularly important in light of the U.N.'s findings that Siri and Alexa's voices reinforce gender stereotypes and biases.

10. Microsoft's digital personal assistant Cortana is named after a character from the popular video game Halo.

Jen Taylor, the same voice actress who portrayed Cortana in Halo, also voices the automated assistant.

11. Researchers are experimenting with "improv chatbots".

Computer science researchers at the University of Southern California have added improv dialogue to chatbots to produce entertaining conversations.

The SPOLIN project consists of over 65,000 pairs of "yes-and" forms of dialogue. "Yes-and" is a fundamental creative exercise in improv that encourages participants to build onto previous ideas.

Researchers fed data from a popular improv podcast to SpolinBot to script funny and wacky responses. The researchers presented their progress at the Association of Computational Linguistics conference in July 2020.

Fed data from a popular podcast about improv

12. When it comes to context, animals are smarter than chatbots

According to Colin Allen, an expert on ethics in artificial intelligence at the University of Indiana, chatbots still have a lot of learning left to do. And animal behavior can help chatbots become smarter.

To fully understand a query and return an accurate answer, digital personal assistants like Siri need to understand context. Allen says that humans tend to adjust their queries whenever chatbots don't understand something. This must change if chatbots are to evolve into the highly skilled assistants we want them to become.

Animals are more skilled at inferring context, relying on the senses, like sight and hearing, to make decisions in the face of uncertainty. Studying this phenomenon can help computer scientists create more adaptive bots.

13. In 2017, two bots at the Facebook Artificial Intelligence Research Lab created their own language.

Bob and Alice started "talking" in a non-human language of their own invention, much to the surprise and confusion of researchers. Dubbed "Facebotlish" by linguistics professor Mark Liberman, this "language" is somewhat unintelligible to humans. The researchers were forced to adjust their model to try to translate and understand the meaning of the conversation.

While this may sound alarming on the surface, researchers determined the conversation was essentially meaningless. (At least, meaningless to us.)

Other computer scientists have observed chatbots create their own language complete with structure and syntax rules. But does this mean chatbots will create complex languages unintelligible to humans? Not quite. As we've already seen, chatbots still have a long road ahead to reach a level of intelligence comparable to the human brain. And this "language" is also text-based, whereas human language also uses speech and body language.

Facts about chatbots: wrapping up

And there you have it — a few fun facts about chatbots that you can use to impress your friends and colleagues. Like this post? Share it with your team!