Hubbry Logo
1 the Road1 the RoadMain
Open search
1 the Road
Community hub
1 the Road
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Contribute something
1 the Road
1 the Road
from Wikipedia

Front cover, 1 the Road

1 the Road is an experimental novel composed by artificial intelligence (AI). Emulating Jack Kerouac's On the Road, Ross Goodwin drove from New York to New Orleans in March 2017 with an AI in a laptop hooked up to various sensors, whose output the AI turned into words that were printed on rolls of receipt paper. The novel was published in 2018 by Jean Boîte Éditions.

Goodwin left the text unedited. Although he felt the prose was "choppy", and contained typographical errors, he wanted to present the machine-generated text verbatim, for future study. The story begins: "It was nine seventeen in the morning, and the house was heavy".[1]

Concept and execution

[edit]

Emulating Jack Kerouac's novel On the Road, Ross Goodwin traveled from New York to New Orleans in March 2017[2] with three sensors, providing real-world input; a surveillance camera mounted on the trunk,[2] trained on the passing scenery; a microphone, picking up conversations inside the car, and additionally the Global Positioning System (GPS), tracking the car's location.[3]

Input from these sources, and the time provided by the computer's internal clock,[1] was fed into a long short-term memory recurrent neural network,[1] which in turn generated sentences on rolls of receipt paper.[3]

The car was a Cadillac; Goodwin explained later he wanted an "authoritative" car (and was unable to get a Ford Crown Victoria), and worried that people might think him a terrorist if they saw the car with its electronics and wires. Google paid part of the cost, having become interested in Goodwin's work at New York University.

Accompanying him were five other people (including his sister and his fiancée), and the Cadillac was followed by a film crew which documented the four-day journey; the documentary was directed by Lewis Rapkin.[2]

Training dataset

[edit]

The training dataset included a sample fiction,[3] consisting of three different text corpora, each with about 20 million words—one with poetry, one with science fiction, and one with "bleak" writing, in Goodwin's words. It had also been fed a data set from Foursquare; the AI recognized locations from Foursquare, and appended commentaries to them.

The conversations captured inside the car were rendered in mutated fashion. The locations provided by the GPS were outputted verbatim, to open the day's writing.[2]

The novel was generated letter by letter.[2] Due to continual input from the GPS and time clock, the novel often mentions the latitude, longitude, and time of day.[1] It was printed unedited and thus is "choppy", according to Goodwin; typos were retained, since he wanted to show the text "in its most raw form".[3]

Goodwin said his main purpose for this novel is to reveal the way machines create words: "In the future when this text becomes more sophisticated it's a warning. If you see patterns like this, it may not have been written by a human".[3]

Reviews

[edit]

Thomas Hornigold, writing for Singularity Hub, concluded that the AI is no Jack Kerouac, but that "you might see, in the odd line, the flickering ghost of something like consciousness, a deeper understanding".[1] Brian Merchant of The Atlantic read the entire novel in one sitting. He could not recognize a coherent plot or story arc, but saw "plenty of pixelated poetry in its ragtag assemblage of modern American imagery. And there are some striking and memorable lines".[2]

Ross Goodwin biography

[edit]

Ross Goodwin, a former ghostwriter for the Obama administration and a creative technologist,[2] has often used neural networks to create poetry and screenplays. Notable works include the short film Sunspring, starring Thomas Middleditch and directed by Goodwin's frequent collaborator Oscar Sharp,[4] and Word.Camera, an 1885 bellows camera that outputs poetry about whatever it is pointed at when the button is pressed.[5] His Master's Thesis at New York University was a project called "Narrated Reality",[6] for which he walked around the city with a backpack containing compass, punch clock, and camera; data from these devices was fed into an LSTM neural network whose output was "weird associative poetry". A year after 1 the Road, Google hired him to work with their Artists and Machine Intelligence project.[2]

Quotes

[edit]
  • "All the time the sun / Is wheeling out of a dark bright ground".[2]
  • "The time was one minute past midnight. But he was the only one who had to sit on his way back. The time was one minute after midnight and the wind was still standing on the counter and the little patch of straw was still still and the street was open".[3]

References

[edit]
[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia
from Grokipedia
1 the Road is an experimental novel generated entirely by an artificial intelligence system during a road trip in the United States in 2017.
The project was conceived and executed by Ross Goodwin, a prominent artificial intelligence creator and "gonzo data scientist" known for blending technology with narrative arts.
To produce the text, Goodwin equipped a black Cadillac with sensors including a surveillance camera, GPS unit, microphone, and clock, which fed real-time environmental data—such as location, time, sounds, and visual descriptions—into a recurrent neural network (LSTM) trained on approximately 60 million words from corpora of poetry, science fiction, and "bleak" literature.
This AI system, running on a laptop, generated the narrative one sentence at a time, outputting surreal and often incoherent prose that printed letter-by-letter on a receipt printer inside the vehicle, mimicking the spontaneous style of Jack Kerouac's On the Road.
The journey began on March 25, 2017, in Brooklyn, New York, and followed a route southward to New Orleans, Louisiana, over four days and evoking the iconic American road trip motif.
The resulting 146-page book, titled 1 the Road, was published in 2018 by Jean Boîte Éditions and marketed as "the first novel written by a machine," highlighting advancements in AI-driven creativity while underscoring the limitations of machine-generated literature, which often lacks coherent plot or human emotional depth.
The work has been recognized for its innovative approach, earning the Doc Lab Award for Digital Storytelling in 2018, and serves as a commentary on surveillance, data, and the intersection of technology and storytelling in contemporary art.

Author and Background

Ross Goodwin's Career

Ross Goodwin is an artist, creative technologist, hacker, and self-described "gonzo data scientist" known for employing machine learning and natural language processing to create new forms of written language. Goodwin began his professional career in digital media and political communications during the 2008 U.S. presidential election, serving as Communications Coordinator for Obama for America in Chicago, where he developed messaging strategies for online and print media targeting Republican supporters. Following Barack Obama's election, Goodwin joined the Obama-Biden Transition Project as staff in Chicago and Washington, D.C., processing resumes and materials for senior government appointments. From 2009 to 2010, he worked as a Presidential Writer at the White House, drafting over 60 official proclamations signed by President Obama, including those for Thanksgiving and Martin Luther King Jr. Day, effectively functioning as a ghostwriter for the administration. After leaving the White House, Goodwin transitioned to roles in financial policy, serving as Special Assistant at the U.S. Department of the Treasury from 2010 to 2011, where he modernized document production processes and drafted reports, and then at the Financial Stability Oversight Council from 2011 to 2012, contributing to operational development. In the mid-2010s, Goodwin shifted toward creative technology and , pursuing graduate studies in interactive at (NYU ITP), where he earned an M.P.S. in 2016 with a on narrated . This period marked his entry into AI-driven narrative experiments, exemplified by his 2015 project word.camera, a device and that uses neural networks and to generate poetic descriptions from photographic inputs in real time, and his 2016 project SUNSPRING, an AI-written screenplay for a short sci-fi film. Building on this, Goodwin established an independent practice focused on AI and language, collaborating on computational storytelling initiatives that bridged technology and literature.

Inspirations and Influences

The primary inspiration for 1 the Road was Jack Kerouac's seminal 1957 novel , which Goodwin sought to emulate through a stream-of-consciousness narrative capturing a cross-country journey. The project's route from to New Orleans directly homaged Kerouac's travels. This literary homage extended to the physical form, as the AI-generated text was printed on a continuous 127-foot receipt paper roll, mirroring Kerouac's use of a 120-foot teletype for his spontaneous composition. Broader influences from literature shaped the project's emphasis on mobility, immediacy, and raw expression, drawing on the movement's rejection of conventional structures in favor of authentic, . Kerouac's techniques of spontaneous , outlined in his 1953 "Essentials of Spontaneous Prose," informed Goodwin's approach by prioritizing undiluted, real-time perception over edited reflection, adapting this to machine-generated output that unfolds in the moment of travel. Additional nods to the road-trip tradition appeared through echoes of authors like and , whose experimental narratives further encouraged Goodwin's boundary-pushing fusion of human and machine storytelling. Technological precedents in AI text generation provided conceptual groundwork, beginning with early experiments such as , Joseph Weizenbaum's 1966 chatbot that simulated conversational patterns through pattern-matching, laying foundational ideas for algorithmic language production. Later milestones like the 1984 book The Policeman's Beard Is Half Constructed, generated by the program, demonstrated AI's potential for creative prose, influencing Goodwin's vision of machines as literary agents despite his acknowledgment that 1 the Road was not the absolute first such work. Modern advancements in neural networks, particularly recurrent models trained on diverse corpora, enabled the project's real-time synthesis of sensory data into narrative, building on these historical efforts to explore AI's role in artistic creation. Goodwin's overarching intent was to produce what he marketed as "the first written by an AI," positioning the project as a deliberate commentary on machine and the blurring lines between authorship and algorithmic output. By framing the car itself as the artist—equipped with sensors to "perceive" and generate text—Goodwin aimed to interrogate whether AI could embody the improvisational essence of beat literature, ultimately viewing the result as an "insect brain that’s learned to write" rather than fully prose. This experiment underscored his role as a "writer of writers," using to extend literary traditions into uncharted territory.

Development Process

Hardware and Sensors

The hardware setup for 1 the Road centered on a rented black , transformed into a "mobile writing machine" to facilitate real-time data capture during the road trip from New York to New Orleans. This vehicle served as the platform for mounting and integrating sensors that provided continuous environmental inputs to the onboard AI system, emulating the sensory experience of a traveler. Key sensors included a white-domed Axis M3007 surveillance camera affixed to the trunk lid, capturing visual data of the road and surroundings; an old GPS unit secured to the roof for precise location tracking; a microphone suspended from the vehicle's ceiling to record in-car audio such as conversations; and the laptop's internal clock for timestamping events. These components were wired directly into a Razer Blade laptop positioned in the back seat, creating a unified system that streamed location, time, audio, and visual data uninterruptedly to feed the neural network's generation process. The setup emphasized portability and minimalism, with the laptop serving as the central hub for data ingestion without additional dedicated hardware. Integration challenges arose from the need for reliable operation over long distances, including issues that necessitated an impromptu purchase at a to sustain the and sensors. Data logging proved demanding due to the continuous high-volume streams—particularly from the camera and GPS—requiring robust buffering to prevent interruptions, though the system ultimately produced over 2,794 feet of printed output without major failures. The visible wiring and equipment also raised practical concerns, such as the risk of attracting unwanted attention from authorities mistaking the rig for suspicious gear.

Software and AI System

The software and AI system for 1 the Road was a custom-built developed by Ross Goodwin, utilizing recurrent neural networks (RNNs) incorporating (LSTM) units to enable sequence prediction and text generation from environmental inputs. This architecture allowed the system to process multimodal data in real time, transforming raw sensor feeds into coherent, if abstract, prose that emulated literary . The model was trained on three distinct text corpora, each containing approximately 20 million words drawn from nearly 200 handpicked books across genres including , , and "bleak" literature, yielding a total training of about 60 million words. To avoid direct imitation of road-trip narratives, the training excluded works by or similar American , instead focusing on diverse stylistic influences to foster emergent, context-driven output. Additional training incorporated location-based data from the Foursquare to enhance place recognition and descriptive commentary. The pipeline began with real-time inputs, including GPS coordinates for location, timestamps from the system clock, audio captures from a for conversational snippets, and visual from a forward-facing camera that recorded images every 20 seconds. Camera images were processed through an image recognition network, which first converted them to representations and then generated a single descriptive sentence per image to serve as a prompt; this, along with other inputs like Foursquare-derived and , was tokenized and embedded as sequential features for the LSTM layers. The RNN-LSTM then predicted subsequent text letter by letter, producing independent sentences at timed intervals that reflected the evolving context of the drive, such as shifting locations or auditory cues. During the four-day road trip from New York to New Orleans in March 2017, the system operated continuously on a , generating unedited text that was immediately printed letter by letter on a receipt printer mounted in the , preserving raw outputs including any inconsistencies or "typos" to highlight the AI's autonomous process. Goodwin monitored the generation and could switch between the three trained corpora mid-trip to adjust stylistic tone if the output became repetitive, ensuring varied narrative flow without post hoc alterations to the core text. This real-time, input-driven approach resulted in over 200,000 words of generated poetry and prose, forming the basis of the published book.

The Road Trip Execution

In March 2017, Ross Goodwin embarked on a four-day from , New York, to New Orleans, , spanning approximately 1,300 miles. The journey began on , with the equipped departing at around 9:17 a.m., capturing real-time environmental to fuel the AI's generation. The route followed a southern path through multiple states, including , , , , , , Georgia, , and , before reaching the destination. Notable stops included , where the vehicle paused at the Eagles Nest Diner, and , for equipment retrieval. While inspired by Jack Kerouac's cross-country travels in , the itinerary was adjusted to maximize diverse sensory inputs for the AI system. During the drive, Goodwin served as both driver and system overseer, positioning his laptop in the back seat to monitor and debug the AI in real time while navigating. Challenges included a traffic halt near Washington, D.C., caused by a right-wing protest, which delayed progress and affected data flow. Power supply issues also arose early on, necessitating an impromptu stop at a convenience store for an additional adapter to sustain the sensors and computing setup. Managing the influx of multimodal data—streaming from the GPS, camera, microphone, and clock—required constant adjustments to ensure seamless real-time processing without overwhelming the onboard hardware. The trip yielded continuous logging of multimodal inputs, including location coordinates, visual feeds, audio recordings, and timestamps, which the transformed into prose printed on receipt paper rolls totaling 2,794 feet in length. This raw data capture formed the experiential backbone of the project's output, emphasizing the journey's immediacy over post-trip editing.

Content and Structure

Generated Narrative

The generated narrative of 1 the Road comprises 143 pages of surreal, non-linear prose created entirely by an system during a four-day from to New Orleans. This AI-produced text adopts a stream-of-consciousness style, evoking the perspective of a sentient as it processes sensory inputs in real time, merging concrete observations of landscapes, weather, and roadside elements with abstract, associative reflections. The prose unfolds through timestamped entries, often brief and fragmented, such as the opening: "It was nine seventeen in the morning, and the house was heavy," which sets a tone of detached, poetic observation. Later passages build on mundane waypoints, evolving into hallucinatory sequences; for instance, descriptions of natural features intensify into evocative imagery like "A in the back was silent and soft and melancholy," blending environmental details with introspective drift. Another example appears in : "Three seconds after midnight. factory, Montgomery. A building in Montgomery to his father's study of this town in the same room where the band was being sent off to the police car. The time was one minute past midnight." Lacking a conventional plot or , the instead forms episodic vignettes tethered to GPS coordinates and temporal markers, capturing the trip's progression through , machine-mediated vignettes rather than sustained .

Stylistic Elements

The stylistic elements of 1 the Road emerge from the AI's data-driven generation process, resulting in a text that blends and through non-sequiturs and unexpected juxtapositions. For instance, the often mixes mundane road observations with philosophical or musings, such as "A patch of water was still looking for some organization, and his lips were falling in vain, a distant ," creating a dreamlike quality that reflects the AI's fragmented interpretation of inputs. Similarly, phrases like "Eagles Nest Diner: a american restaurant in Goldsborough or the Marine Station, a place of seemed to be a man who has been assembled for three days" exemplify this , where everyday landmarks dissolve into illogical assemblies devoid of human logic. Central to the work's themes are and consciousness, as the AI narrates from the perspective of its sensors, questioning the boundaries between human and artificial viewpoints. Entries frequently incorporate GPS coordinates and timestamps, such as "35.415579526 N, -77.999721808 W, at 154.68504432 feet above ," to evoke how the "sees" the world through rather than emotional filters, blurring the line between objective recording and subjective . This exploration manifests in surreal metaphors like "The road was a ribbon of time," which personify environmental elements in ways that probe the emergence of from algorithmic processes. Linguistically, the text exhibits repetitive phrasing inherent to the (RNN) architecture, alongside poetic fragmentation that echoes beat poetry traditions. Repetitions such as "The road stretched. The road waited" occur frequently, comprising about 25% of certain motifs, lending a rhythmic insistence but lacking emotional depth, which underscores the AI's pattern-based output. Fragmented sentences, like "I somewhat when i’m on why i didn’t get hurt yeah my car is an every down i know?," further contribute to a loose, grammatically unconventional that produces both profound and absurd effects, resembling a long, stream-of-consciousness poem. Comparisons to human-authored works highlight parallels to Jack Kerouac's , particularly in its jazz-like rhythm and motif, yet with a data-driven detachment that strips away personal introspection. The AI's real-time prose generation mimics Kerouac's spontaneous scroll method— spans 127 feet of continuous text—but replaces humanistic vitality with machinic observation, as in poetic lines like "the showed a past that already had hair from the side of the track." This detachment fosters a unique aesthetic, where echoes of beat spontaneity serve algorithmic echoes rather than .

Publication and Reception

Publication Details

1 the Road was initially published in 2018 by Jean Boîte Éditions in Paris as a bilingual English-French edition in hardcover format, comprising 146 pages. The book featured an introduction by Kenric McDowell and graphic design by Joanna Starck, with the ISBN 978-2-36568-027-1. The publication was marketed prominently as "the first novel written by a machine," highlighting the AI's authorship while downplaying Ross Goodwin's facilitating role, positioning it as a groundbreaking experiment in machine-generated literature. Distribution involved a limited first print run through the publisher, with digital excerpts made available online for broader access. Later editions reached U.S. audiences via platforms like Amazon, expanding its availability beyond the initial European release. Accompanying the book was a titled 1 the Road (2018), directed by Ross Goodwin, which documented the project's creation and road trip execution; it premiered at the (IDFA) and won the IDFA DocLab Award for Digital Storytelling.

Critical Reviews and Impact

Upon its release, 1 the Road received praise for its innovative approach to machine-generated storytelling. In a 2018 article, The Atlantic described the work as compelling, likening it to a Google Street View car narrating its own cross-country journey and highlighting its fresh take on automated narrative generation. Similarly, a review in BOMB Magazine that year commended its as captivating and poetic, noting unexpected phrases that evoke an insect-like consciousness attempting human expression, and positioning it as a worthwhile read in the tradition of experimental road literature. Critiques were mixed, with some reviewers pointing to the AI's limitations in producing coherent narratives. As of November 2025, the book holds an average rating of 3.2 out of 5 on based on 4 reviews, where users frequently noted its fragmented and often nonsensical passages as detracting from readability, though others appreciated its experimental charm. Despite these shortcomings, scholars have viewed 1 the Road as a milestone in procedural literature, demonstrating how algorithmic processes can generate extended narrative forms from real-time sensory data. The project significantly impacted discussions on AI authorship and in creative fields. It prompted debates about and creativity in machine-generated works, as explored in legal analyses questioning attribution for AI outputs like this . 1 the Road influenced subsequent AI literary experiments, including ongoing projects by creator Ross Goodwin that build on narration, and contributed to broader explorations of human-AI collaboration in writing. It was recognized with the IDFA DocLab Award for in 2018 and listed among the Breakthroughs in Storytelling by the Digital Dozen awards. Culturally, 1 the Road left a legacy through its exhibition at the 2018 (IDFA), where it was showcased as an immersive example of AI-driven nonfiction storytelling. The work has fueled ongoing AI ethics conversations in , emphasizing the tension between technological innovation and artistic intentionality in procedural and forms.

References

Add your contribution
Related Hubs
Contribute something
User Avatar
No comments yet.