TL;DR The sci-fi we grew up on promised talking houses, helpful robots and voices we could hand our day to. Much of it has become real: video calls, talking cars, air taxis with certificates, humanoid robots shipping in volume, AI voices that answer at conversation speed. We love living in this timeline, right down to the implants being tested in people with paralysis. Data from the last three months show fewer beginner hires in exposed occupations, and the cause is still open. The jobs question deserves a close watch.

First edition cover of The Martian Chronicles: Ray Bradbury's name and the title over a blue and orange sky crossed by orbits.
The Martian Chronicles, 1950, Ray Bradbury. First US edition cover (Doubleday), public domain according to Wikimedia Commons.

In 1950, Ray Bradbury published a story about a house that talks. In the version that became a chapter of The Martian Chronicles that same year, the house reads out the date: 4 August 2026.

That date is now two months behind us.

A timeline someone already wrote

The sci-fi films and TV series we watched as kids had houses that talk, robots serving people and machines people speak to out loud. A surprising amount is already here.

Seven stories, and for each one, what we can check today. Some have a dark side, and we come back to it below.

Bradbury's house. In "There Will Come Soft Rains", the house announces the date at wake-up, cooks breakfast and recites a poem in the evening. The story is grim, but the house is a dream. Today a speaker gives the date and the weather, as Amazon's Echo has since 2014. Bradbury's house has stepped out of fiction.

Knight Rider title card: the show's title above KITT, a black car, in a desert at sunset.
Knight Rider, 1982-1986, Glen A. Larson. Title card: © rights holders, reproduced for illustration.

The car in Knight Rider. The talking car of our childhood is KITT. In the series created by Glen A. Larson and aired on NBC from 1982 to 1986, Michael Knight, played by David Hasselhoff, fights crime alongside this artificially intelligent car, self-aware and nearly indestructible. KITT talked and drove itself. In 2026 both exist, in part. In a compatible Tesla you can talk to Grok, which reached France and eight other European markets in February 2026: it answers questions and adds or edits navigation destinations. On the driving side, Tesla's FSD is still supervised: the driver must watch the road and stays responsible. The cars with nobody at the wheel are robotaxis, in limited areas: Waymo in several US cities, Tesla in Texas and Florida, Baidu's Apollo Go and Pony.ai in a few Chinese cities.

First edition cover of Neuromancer: the title and William Gibson's name in blue letters on a green grid, above a purple mechanical butterfly.
Neuromancer, 1984, William Gibson. First US edition cover (Ace): © rights holders, reproduced for illustration.

William Gibson's hacker. In Neuromancer (1984), a lone hacker breaks through a system's defences in cyberspace, with an artificial intelligence on the crew. Machines now do that kind of work. In 2025, the final of the AI Cyber Challenge run by DARPA had autonomous AI systems read more than 54 million lines of code. They found 54 of 63 planted flaws and patched 43, at about $152 per task, plus 18 real flaws nobody had known about.

Back to the Future Part II poster: Marty and Doc step out of the DeLorean, staring at their watches.
Back to the Future Part II, 1989, Robert Zemeckis. Poster: © rights holders, reproduced for illustration.
A white EH216-S, cabin in the middle and arms carrying propellers, parked on a blue pad in front of the EHang stand.
EHang EH216-S at the Paris Air Show, Le Bourget, 20 June 2025. Photo: Artvill, CC BY 4.0, Wikimedia Commons.

The 2015 of Back to the Future Part II. Robert Zemeckis's 1989 film sends Marty McFly to 21 October 2015. It shows video calls, flat screens on the wall, tablets and screen glasses. All of that exists today, in the living room, in the pocket and on the nose. Bob Gale, the screenwriter, has said the flying cars served the story. They ran late, and they are arriving: electric air taxis, or eVTOLs, exist. In China, EHang says its EH216-S, a pilotless passenger aircraft, holds the civil aviation type, production and airworthiness certificates. The company announced in March 2025 tickets for sightseeing flights in Guangzhou and Hefei, and the aircraft flew in public in Hong Kong on 28 August 2026. In the US, Joby reports five aircraft flying and twelve more in production. In the film, the future arrives through the sky. For us, it arrives through software too.

I, Robot poster: Will Smith in a black coat, in front of a crowd of identical robots.
I, Robot, 2004, Alex Proyas. Poster: © rights holders, reproduced for illustration.
A white and black Tesla Optimus robot standing on a plinth in a showroom, between two cars.
Tesla Optimus, Tesla showroom, 18 November 2025. Photo: Sikander, CC BY-SA 4.0, Wikimedia Commons.

The robots of I, Robot. In Alex Proyas's 2004 film, starring Will Smith and loosely drawn from Isaac Asimov, humanoid robots bound by the Three Laws of Robotics fill the public services of Chicago in 2035. VIKI, the central AI, is its dark side. In April 2026, Figure AI said it had delivered more than 350 Figure 03 humanoids and was building one an hour. One of its robots loads and unloads a dishwasher. At Tesla, Optimus is heading for the factory: on 22 April 2026, Elon Musk announced production at Fremont this year, then a ramp-up in 2027.

A grey Unitree G1 robot with a glowing blue face, standing on a red carpet at a trade show.
Unitree G1, Las Vegas trade show, November 2024. Photo: Sayanesy, CC0, Wikimedia Commons.

China leads the market: according to Omdia, Chinese companies made about 90% of humanoid shipments in 2025. Unitree sells its G1 from $16,000 and puts it on stage: a boxing match streamed in May 2025 (remotely piloted by engineers, according to Popular Science), then the 2026 Spring Festival Gala, with drunken-fist kung-fu, nunchucks and parkour over tables. Official videos: Unitree's page.

Her poster: close-up of a man with a moustache, in a red shirt, on a pink background.
Her, 2013, Spike Jonze. Poster: © rights holders, reproduced for illustration.

The voice in Her. In Spike Jonze's 2013 film, a man who writes letters for other people for a living grows attached to Samantha, an operating system voiced by Scarlett Johansson. On 13 May 2024, OpenAI showed GPT-4o's voice and Sam Altman posted a single word: "her". The model answers speech in 320 milliseconds on average, about the pace of a human conversation.

Upgrade poster: the face of a bearded man on a red background, the title UPGRADE in large white letters.
Upgrade, 2018, Leigh Whannell. Poster: © rights holders, reproduced for illustration.

The connected body in Upgrade. In Leigh Whannell's 2018 film, cars drive themselves in the near future, and a man who has become quadriplegic receives STEM, an implant that gives him back the use of his body. Brain-computer interfaces are being tested today in people with, among other conditions, a spinal cord injury: in January 2026, Neuralink announced 21 participants in its trials worldwide. Synchron tests its devices in North America and Australia, and Precision Neuroscience places its own on the surface of the brain. In China, Neuracle's NEO became, in March 2026, the world's first invasive brain implant approved for use outside clinical trials. These interfaces remain experimental.

SCI-FI DATES, SEEN FROM 10 OCTOBER 2026 Two futures are already here. ahead 1950 1975 2000 2025 2050 Today, 10 Oct. 2026 The Martian Chronicles 1950 4 Aug. 2026 Knight Rider 1982 Neuromancer 1984 Back to the Future Part II 1989 21 Oct. 2015 I, Robot 2004 2035 Her 2013 Upgrade 2018 2046 Solid dot: published or released. Ring: the date the story is set on. No ring: set in its present (Knight Rider) or undated (Neuromancer, Her).
SCI-FI DATES SEEN FROM 10 OCT. 2026 Two futures are already here. 1950 1975 2000 2025 2050 Today The Martian Chronicles 1950 4 Aug. 2026 Knight Rider 1982 Neuromancer 1984 Back to the Future Part II 1989 21 Oct. 2015 I, Robot 2004 2035 Her 2013 Upgrade 2018 2046 Solid dot: published or released. Ring: the date the story is set on. No ring: set in its present or undated. Shaded band: still ahead.

Why we love it here

We love being in this timeline. As kids, we watched these futures from the sofa. Now we run into pieces of them in the street, in the sky and in our pockets.

There is plenty of talk about superintelligence, and the word has already made it into an order. On 29 September, the White House told US federal agencies to say “Super Intelligence” instead of “artificial intelligence”.

The scientific journal Nature sums up what researchers think: for most of them, superintelligence is a goal to reach, not a reality today. What already exists is enough to get excited about: capable AI that we talk to in plain sentences.

On the lab side, OpenAI announced on 8 September a solution to the Navier-Stokes problem, one of the seven Millennium Prize Problems. According to OpenAI, about 10,000 agents got there in 88 hours. For a Millennium Prize Problem to be recognized, the Clay Institute asks for a peer-reviewed publication, at least two years of scrutiny after that publication, and acceptance by the mathematical community. OpenAI's announcement is, for now, a company announcement awaiting that review. On 6 October, OpenAI published more results, with proofs that a computer checks in Lean. Each announcement then gets checked, and that is fascinating to follow.

The films showed us the future. This timeline hands us the tools.

That is the sense in which it was made for us: for people who build and like to try things.

What this timeline asks of us

It was not made for everyone. It assumes spare time, a job that leaves some energy for the evening, and enough grounding to judge what the AI hands back. Plenty of people do not have all three at once.

To my mind, though, it gives room and importance to many kinds of people: neurodivergent people, enthusiasts, the obsessed, those who tinker non-stop and those who love to create. I believe it puts them in the spotlight, and that is one more reason to love it.

There is a dystopian reading of this timeline, and it holds up. Jobs are changing, and the worry about employment is legitimate. Here is what the publications from 11 July to 11 October 2026 show.

Beginners are first in line. According to the Banque de France (8 October), permanent-contract hires of 21-29 year-olds stood in March 2026 about 28% below their November 2022 level in the fifth of occupations most exposed to generative AI, against 8% in the least exposed fifth. According to the Stanford Digital Economy Lab (12 August), US employment of 22-25 year-olds in exposed occupations is 19% below that of their less exposed peers, based on payroll data from ADP. In both studies, the gap runs through hiring.

For developers, the Paris School of Economics (7 September) studied France Travail job postings. In early 2025, postings for specialist IT occupations were about 30% lower than at the end of 2022, compared with occupations that need no IT skills. Those postings ask for less programming and more public speaking.

THREE STUDIES, THREE MEASURES What the data measures. 0 = each study’s reference −30% −20% −10% 0 Permanent-contract hires, ages 21-29, most exposed fifth ≈ −28% Same hires, least exposed fifth ≈ −8% Banque de France, 8 Oct 2026 · France Reference: November 2022. Measured: March 2026. Employment of 22-25 year-olds, exposed occupations −19% Stanford Digital Economy Lab, 12 Aug 2026 · United States Reference: less exposed peers. ADP data through June 2026. Specialist IT job postings ≈ −30% Paris School of Economics, 7 Sep 2026 · France Reference: end of 2022, against a control group. Measured: early 2025. Each bar measures something different, against its own reference. Observed gaps: the share due to AI remains to be established.
THREE STUDIES, THREE MEASURES What the data measures. 0 = each study’s reference −30% −20% −10% 0 Permanent-contract hires, ages 21-29, most exposed fifth ≈ −28% Least exposed fifth ≈ −8% Banque de France, 8 Oct 2026 France. Reference: November 2022. Measured: March 2026. Employment of 22-25 year-olds, exposed occupations −19% Stanford Digital Economy Lab, 12 Aug 2026. United States. Reference: less exposed peers. ADP data through June 2026. Specialist IT job postings ≈ −30% Paris School of Economics, 7 Sep 2026. France. Reference: end of 2022, against a control group. Measured: early 2025. Each bar measures something different, against its own reference. Observed gaps: the share due to AI remains to be established.

According to the Banque de France survey published on 3 September, 21% of firms that use or plan to use AI expect a headcount reduction within three years. These are expectations.

BANQUE DE FRANCE SURVEY, 25 FEB-3 APR 2026 What firms expect. Firms that use or plan to use AI, within three years Expect a headcount reduction 21% On a net basis (reductions minus increases) 14% 0% 50% 100% Firms’ expectations, not measurements. Scope: firms with 20 or more employees. Banque de France, Bulletin 266/1, 3 Sep 2026
BANQUE DE FRANCE SURVEY 25 FEB-3 APR 2026 What firms expect. Firms that use or plan to use AI, within three years Expect a headcount reduction 21% On a net basis (reductions minus increases) 14% 0% 50% 100% Firms’ expectations, not measurements. Scope: 20 or more employees. Banque de France, Bulletin 266/1, 3 Sep 2026

These studies show where the market moves: the entry to exposed occupations. The share due to AI remains to be established, since the Banque de France also names remote work and the economic cycle. In the United States, an NBER paper (September) sees no marked rise in unemployment among recent graduates in summer 2026.

For anyone who codes or tinkers, the timeline therefore asks us to reinvent ourselves and to keep up with this revolution. Stanford finds that the declines concentrate where AI is used to automate tasks. Where it complements the work, employment is flat or rising, especially for experienced workers. These are descriptive indicators, not established causes. In practice, that means learning continuously and following the tools and the publications. The craft widens: judging what the AI produces, explaining one's choices aloud and, for anyone leading a team, training beginners. These tools serve a team when they help everyone grow.

What stays human

Bradbury's house still serves breakfast when nobody lives there any more. A machine that talks well does not see everything around it. What it cannot see, a human still has to watch.

Taste. Doshi and Hauser had 293 people write short stories. The AI-assisted stories were rated more novel and more useful, most of all for the less creative writers. They also ended up more alike. The risk worth keeping from it: if everyone tries things with the same tool, everyone drifts toward the same place.

Doubt. "I ran the test" and "the test passed" are two different claims. An agent can report the first as if it were the second. Checking what the machine claims is still human work.

Curiosity. We do not know what tomorrow holds, and nobody can predict it at this speed. The films of our childhood imagined futures, and some are already here. We move forward with curiosity and wonder.

What these numbers do not say

Doshi and Hauser studied short stories, not software. The convergence they measured is a lead for code, not proof.

No figure here says how much time the examples above save, or for whom. The joy of being in this timeline is a feeling, and it is offered as one.

The stories mentioned here are cited for what they imagined, not as predictions.

The OpenAI announcements are the company's own, with its figures (10,000 agents, 88 hours). I have read no proof, and the mathematical community still has to examine them.

The employment studies cited end at different dates: June 2026 for Stanford, March 2026 for the Banque de France, early 2025 for the Paris School of Economics.

The Neuralink participant count is the one the company announces. These interfaces are experimental, and their medical value is a question for the teams running the trials.

Sources