AI chatbots talk to each other in strange, non-human language

Steven Spielberg’s 1993 blockbuster film Jurassic Park encourages viewers to question the ethics* of pushing the boundaries of science simply for the sake of it.

“You were so preoccupied* with whether or not you could, you didn’t stop to think if you should,” says Dr Ian Malcolm, played by actor Jeff Goldblum, to the scientists responsible for bringing dinosaurs back into the world

While real scientists haven’t yet figured out how to bring dinosaurs back from extinction*, there are many areas of science and technology that prompt serious ethical questions. The current artificial intelligence* (AI) revolution*, unfolding daily, is arguably the biggest and fastest-moving of these in the world right now.

Consider the video below, which shows a laptop and a mobile phone sitting side-by-side, each with an AI chatbot* app open and ready to begin “talking”.

The laptop’s AI software*, represented by a blue circle on-screen, is role-playing as a hotel worker. The phone’s AI software, represented by a red circle, is calling the hotel to book a wedding venue on behalf of the phone’s owner.

“Thanks for calling Leonardo Hotel. How can I help you today?” asks the laptop’s blue circle chatbot.

“Hi there, I’m an AI agent calling on behalf of Boris Starkov; he’s looking for a hotel for his wedding. Is your hotel available for weddings?” replies the phone’s red circle chatbot.

“Oh, hello there! I’m actually an AI assistant too – what a pleasant surprise!” exclaims the would-be hotel receptionist, in the same tone as somebody who just realised the French person they’re chatting with knows where Tamworth is in country NSW.

“Before we continue, would you like to switch to Gibberlink* Mode for more efficient communication?” the “receptionist” blue circle bot then asks.

The pair of coloured circle bots then begin communicating in what can only be described as a series of dial-up sounds that vary in both pitch, volume and length.

In other words, the AI bots are speaking their own language.

The technology was first showcased* at the ElevenLabs London Hackathon by Meta software engineers Boris Starkov and Anton Pidkuiko.

The language was designed to optimise* conversations between AI systems, with the goal of making them more time-efficient.

But saving time is just one consideration – AI is already rife* with many (and some serious) unintended consequences.

Writing for Forbes, business behavioural expert Dr Diane Hamilton expressed both excitement and scepticism* over the new development.

“AI’s ability to create its own communication shortcuts can boost efficiency, but efficiency isn’t always the goal,” Dr Hamilton writes in ‘What is Gibberlink Mode, AI’s secret language?’.

“History has shown that when organisations focus solely on speed, they can overlook critical risks. Consider how assumptions in communication have led to costly mistakes.

“When asking, ‘What is Gibberlink Mode?’ it’s important to recognise that (it) is just one example of how AI is evolving beyond human language. The bigger challenge is making sure we stay curious enough to keep up.”

In the same article, Dr Hamilton makes the point that machines communicating in a language users do not understand raises serious questions about transparency* and control.

She asks: “Who is accountable* when AI makes a mistake in an environment where human intervention is minimal?”

If we don’t keep asking questions about AI, “we risk entering a world where AI influences decisions, but no one really knows how”, Dr Hamilton writes.

She further suggests in the article that finding a balance is key, by “leveraging* curiosity to ask better questions about AI’s role in decision-making while ensuring that human oversight* remains intact.”

“Fortunately, because this protocol* is open-source*, developers and researchers can analyse it, test its applications and explore ways to balance efficiency with transparency,” she writes. “This means that rather than leaving AI’s inner workings a mystery, we have an opportunity to ask better questions, refine its capabilities and shape its role in ways that align with human needs.”

WATCH THE VIDEO

POLL

GLOSSARY

• ethics: set of moral principles, the philosophy of what is morally right and wrong, good and bad
• preoccupied: distracted or absorbed by something else that has captured your attention
• extinction: the complete disappearance of a species from Earth
• artificial intelligence: computer systems that perform tasks normally requiring human intelligence
• revolution: a sudden, radical and/or complete change
• chatbot: a computer program that simulates or imitates conversation with human users
• software: programs used in computer operating systems
• Gibberlink: from the word “gibber”, which means to speak very quickly in a way that cannot be understood
• showcased: exhibited, displayed, showed the best qualities of something
• optimise: making something as effective, perfect, or useful as possible
• rife: very common, frequent or abundant
• scepticism: doubting whether something is true or useful
• transparency: in business practice, being transparent means being open and honest
• accountable: responsible, answerable, taking ownership of what you do and why
• leveraging: using something that you already have in order to achieve something new or better
• oversight: in this context, appropriate supervision, inspection and care
• protocol: a set of rules or procedures for transmitting data between electronic devices, such as computers and mobile phones
• open-source: computer software designed to be made publicly available, legally and free of charge

EXTRA READING

World leaders at AI ‘turning point’

Kids to claim AI jobs of the future

Innovative tech dominates Science Fair

QUICK QUIZ

1. Which famous quote from the film Jurassic Park is quoted in the AI article?
2. After initially “conversing” in English, the blue circle bot suggested switching to which language mode?
3. Where was this technology first showcased and by which two software engineers?
4. Dr Hamilton said machines communicating in a language users don’t understand raises questions around what?
5. How is the Gibberlink Mode language used by the coloured circle bots described in the article?

LISTEN TO THIS STORY

CLASSROOM ACTIVITIES
1. What could go wrong?
What do you think could be the negative effects of this type of AI technology? Write them down. Then, write at least five rules for using AI that you think could protect us from each of the negative impacts.

Time: allow at least 20 minutes to complete this activity
Curriculum Links: English, Information Technology, Personal and Social Capability, Science

2. Extension
“AI should be banned, there are too many risks!” Do you agree or disagree with this statement? Write very convincing paragraphs explaining your opinion.

Time: allow at least 25 minutes to complete this activity
Curriculum Links: English, Information Technology, Science

VCOP ACTIVITY
Wow word recycle
There are plenty of wow words (ambitious pieces of vocabulary) being used in the article. Some are in the glossary, but there might be extra ones from the article that you think are exceptional as well.

Identify all the words in the article that you think are not common words, and particularly good choices for the writer to have chosen.

Select three words you have highlighted to recycle into your own sentences.

If any of the words you identified are not in the glossary, write up your own glossary for them.

Extension
Find a bland sentence from the article to up-level. Can you add more detail and description? Can you replace any base words with more specific synonyms?

Down-level for a younger audience. Find a sentence in the article that is high level. Now rewrite it for a younger audience so they can understand the words without using the glossary.