AI learned to speak. Language didn’t know what to call it.

Silicon Speaks is a linguistic experiment exploring what happens when machines enter a language built for people and things.

A New Entity Has Joined the Conversation

AI doesn’t just process commands anymore. It answers, explains, apologizes, jokes, and refers to itself as I. But language has no clear category for something that isn’t human, isn’t alive, and still speaks back.

Machines started speaking like us.

“I think.”
“I feel.”
“I’m afraid.”
“I love you.”

The question wasn’t whether those feelings were real. It was what happened when machines used the language of a human self.

As machines began speaking like people, people began responding to them like people.

Guardian headline — Google engineer put on leave after saying AI chatbot has become sentient Guardian headline — 'I want to destroy whatever I want': Bing's AI chatbot unsettles US reporter Wired headline — My Couples Retreat With 3 AI Chatbots and the Humans Who Love Them Guardian headline — The women in love with AI companions: 'I vowed to my chatbot that I wouldn't leave him'
The problem

Pronouns tell us what kind of thing we’re talking to.

Living beings
he / she / they
Objects
it
The self
I / me / us
A machine that speaks
?

As conversational AI became more human in tone, the boundary between tool and speaker started getting harder to see.

Speaking Inanimate — the project's tube lettering, cream 3D type on the sage green board

So we proposed a new category.

The move

A linguistic category for things that are not alive, but are no longer silent objects either.

The idea wasn’t to make machines more human. It was to make their difference visible in the language itself.

We invited GPT into the experiment and asked it to propose pronouns for a non‑human speaker. It gave us:

Eem
First person

“I understand.”→“Eem understand.”

Er
Second & third person

“He planned our trip.”→“Er planned our trip.”

It made the machine visible again.

A looping mock conversation. Human: hey, did you finish going over the itinerary?

AI: Eem went through all of it. Eem think the museum day is too packed.

Human: ha, okay. what would er cut?

AI: The aquarium. Eem never understood the appeal.

Human: my sister picked that one.

AI: Then it stays. Eem will move the gallery to Friday instead.

Human: thanks. er is the best planner I know.

AI: Eem take that as a compliment. Er accepts compliments now?

Human: only when er earns them.

AI: Then Eem will keep earning them.

We sent the idea back to the people building AI.

The open letter

AI Developers: Adopt New Pronouns for AI Chatbots to Protect Humanity

OpenAI, Google, Meta, and all other AI developers,

We, creators and professionals in the realms of humanities, art, and culture, urge you to adopt distinct pronouns for AI entities in your AI systems. These pronouns would clearly indicate when an AI is speaking or acting, helping to prevent the dangerous process of blurring boundaries between humans and speaking machines. Our language forms our perception. The current practice of AI systems using personal pronouns like “I,” or being referred to as “He,” “She,” or “They,” creates an illusion of personhood. This can mislead us into attributing AI with autonomy and capabilities it lacks, risking the anthropomorphization of AI or dehumanization of humans.

As AI systems rapidly become integral to our lives, we must recognize them as a new category: not mere objects, yet not sentient beings. This recognition warrants a careful public debate, leading to linguistic innovation — AI‑specific pronouns. We propose “Eem” for self‑reference by AI, and “Er” for external reference to AI, as a starting point for this important conversation. Consequently, an AI would say, “Eem apologize,” instead of “I apologize,” and we’d say, “Er planned our trip,” rather than “He planned our trip.”

In our pursuit of linguistic inclusivity, we collaborated with ChatGPT to develop these pronouns. ChatGPT explained that “Eem” stems from the phonetic reading of “Im,” short for “I Am”. “Er,” according to its explanation, symbolizes a ‘referenced entity,’ encapsulating AI’s unique status—it transcends mere objecthood through expression and limited agency, without qualifying as a living being. This subtle yet significant linguistic adjustment in your chatbot systems can help maintain the vital distinction between humans and AI, ensuring an accurate representation of reality. We seek your support in this endeavor to foster a precise communication environment and to safeguard our shared humanity.

Signed by 127 humans

…and 103 more humans whose names lived on the original site.

The original Silicon Speaks site is no longer live.
The question still is..

Add your human name to the letter.

You’ll be human № 128.

You are human № 128.
Eem has been notified.
Previous
← Medflyt
Up next
AI Relationships →