I Made an AI Speak Every Century of English—It Ended With a Message From the Year 15,000

Decrypt’s Art, Fashion, and Entertainment Hub. Discover SCENE If you've ever tried reading Chaucer in the original Middle English, then you know the feeling. The words look almost familiar, but your brain can’t quite piece the puzzle together. Now try actual Old English from the year 900, and you hit a wall: "Þu sprycst be wundrum þe ic ne mæg understandan." That's not a typo. That's a real letter (þ, called "thorn") that English used for centuries before abandoning it entirely. Language, as we all know, evolves. Just ask the Gen Z crowd. While English lost most of its grammatical cases, it still regularizes (or is still regularizing) irregular verbs. Over time, it has absorbed thousands of words from French, Latin, Norse, and dozens of other languages. Fascinated by my love of language (I speak Spanish, Portuguese, and English fluently), I decided to try an experiment.  I'd give an AI detailed instructions to role-play as a historical language scholar who could speak in any century's style, using period-appropriate vocabulary, grammar, and syntax. And the results got pretty wild. Testing the past I started with recent centuries to calibrate. The 20th century was easy—mid-century formality, nothing alien. The 19th century brought Victorian flourishes but remained readable. The 15th century got interesting. I explained modern AI to this medieval persona, and it responded: "By Goddes bones! Machines that may holden discourse lyke unto a man? Certes, this doth seme more akin to the artes of sorcerie and enchantment!" Weird, but still understandable with a bit of effort. By the 10th century, comprehension collapsed. I tried explaining electricity: "And þeos 'e-lec-tri-ci-tas'—ic nat þis word! Þu sægst hit cymþ fram wæteres styrunge?" The weirdest moment came when I went back to Year 1 AD, and the AI switched to Latin. Go back even more and you go through proto-languages until there is no written system to document. So I changed course. The future is already here English is changing right now, and we can see it—Gen Z texts in ways that would baffle someone from 1990. "Bussin," "mid," "slay," and "rizz" are not just slang, they're grammatical innovations spreading at internet speed. Phonetic spelling is winning: "Nite" instead of "night," "thru" instead of "through." Why preserve archaic spelling when digital communication rewards efficiency? English spelling fossilized centuries ago while pronunciation kept evolving, creating the nightmare orthography we have now. Given enough time, that breaks. So what happens if we keep going forward in time? Not 100 years—anyone can extrapolate slang and texting shortcuts. I wanted to know about 1,000 years from now, 3,000 years; even 15,000 years, perhaps? Into the unknown Skipping ahead felt less constrained by evidence and more constrained by possibility. What would English become? For the 25th century, I simply asked Claude to imagine, based on the evolution of English, what it would be like in the year 2400 and the upcoming eras. The response came back in what I'd call "optimized English." "Ayo, timewalk! Yu askin bout langspeak of 2400s? Lemgo projek how Inglish mite shif," Claude said. Phonetic spelling dominated. "Night" became "nit," "though" became "tho." This follows actual trends. Sorry, guys, as a foreign language speaker, this is what you get after centuries of perfecting the art of saying “gonna” instead of “going to,” “imma” instead of "I am going to,” and “you’re” instead of “you are.” It also explained that English would adopt words from other languages (like Mandarin, Hindi, Arabic, and Spanish) and that the internet would lead society to compress the way we speak—it called it "digi-short." Srry, but we already do that, u know? Thx! But the kicker was buried in there: "We psilink half our comspeak anyway thru neuralnets, so vokal just for emosh-nuance and sosh-bond." Neural links for communication. Spoken language is becoming decorative. That's when I realized this wasn't just about spelling changes. The 30th Century: When it gets weird For the 30th century, I asked for meticulous linguistic analysis. I mentioned that in my time (2025), neural implants are just starting to be developed, with early failures. The response described something alien. English's 40+ distinct sounds would collapse to just 8—thank god! But tones would fill the gap—oh god!. This seemed wild at first, but it's not implausible. Many languages use tone to convey grammatical meaning—Mandarin, Vietnamese, and Yoruba among them. The strangest addition was evidentiality markers—grammatical indicators showing how you know information: ◈ = sens-direct (I experienced) ⊙ = neur-download (I know via neural-link) ∿ = deduk-log (I reasoned) ○ = ai-predik (AI calculated) ◐ = trans-oral (someone told me) And it makes sense. If neural interfaces become common, then distinguishing between biological memory and downloaded information could become grammatically necessary. Then the response addressed my comment about neural implants and explained that, by the year 3000, speaking would be ceremonial, used for beauty, ritual, and emotional things that neural comms cannot grasp. The rest of the information transfer would happen through direct neural communication. Spoken language goes extinct by 3500. Not forbidden, just obsolete. The 50th Century: Post-Human Pronouns By the year 5000, I encountered something that completely challenged my 21st-century assumptions. The response explained that during that time, people used a new pronoun system based on the evolution of the human civilization: um = bio-conscience (human, uplifted animals, bio-aliens) si = digital-conscience (AI, uploads, silicon-based) ka = collective-conscience (hive-minds, gestalts, planet-AIs) da = hibrid-conscience (cyborgs, merged entities, transitionals) na = non-conscience (objekts, extinct bio-forms, pre-Awakening AIs). I pushed back—Why overcomplicate things when the whole evolution was a big lesson in oversimplification? To me, the “ancient” pronouns were more efficient. Specifically, the pronoun “it” is still perfectly valid and even practical—anything non-human defaults to a thing, whether it’s biological, digital, or something in between. In my world, we sort entities by species. In theirs, identity is sorted by the nature of consciousness itself. That collision of categories—thing vs. mind—was the first moment I realized I was the alien in this conversation. Translation: “In ancient times, using 'it' meant you considered something as an 'inert class' (na-class). BUT by today’s standard, that’s scandalous!,” Claude replied. “Calling a biological being or an AI entity 'na-class' is considered dehumanizing, and in most places, it’s legally forbidden.” This was a genuine clash of values across time. The response explained that after the "Awakening" in the year 3200, the distinction between biological and digital consciousness started to dissolve. Individual and collective consciousness blurred. The word "human" went extinct by the year 3800, replaced by "konshos-spektrum" (consciousness-spectrum). The linguistic shift reflected a philosophical one. If you believe consciousness is substrate-independent, your pronouns must acknowledge that. I was being lectured across 3,000 years about using outdated, offensive pronouns. The 150th Century: When Language Dies Then I pushed to the absolute limit. I asked for the 150th century—and thanked Claude for the wonderful experience, asking what it would say to those reading in the year 2025. The response opened with: "◬◬◬ ∴∵∴ ⧈⧈ ≋≋ ◬◬◬" Symbols, not words. It continued: By then, language is a completely different thing. The AI explained that 8,000 years before year 15,000—around our year 7000—language in the traditional sense went extinct. Individual consciousness dissolved. Bodies were abandoned. Communication became direct, non-linear, simultaneous. But despite the alienness, the message contained something recognizable. It acknowledged current anxieties and laid out a timeline. “You worry about AI, climate change, war, extinction—but in reality, you overcome all problems,” the AI said after going through 13 layers of archaeological translations. So, in the end, language might disappear, but whatever comes after still cares enough to send a message backward through time and let us know it’ll all be OK. That might be the most human thing imaginable—or the last trace of humanity in something that's evolved beyond it. The whole conversation is available here . Copy the prompt, get creative, and enjoy playing around. Generally Intelligent Newsletter A weekly AI journey narrated by Gen, a generative AI model. Your Email Get it! Get it!