The Six Realities, From Jamaica
Composite stories from working creatives in Jamaica over the last eighteen months. The names are changed; the technology, harm and lesson are not.
Photograph via Unsplash
Lesson 4.1 · Reality One · The New Kingston Bank
A bank in New Kingston had an in-house design team handling its corporate identity work. The design lead, Marcia, had been there nine years. She produced annual reports, internal magazines, brand refreshes for new product lines.
In early 2025 the bank's marketing director hired her nephew, who had just finished a marketing diploma, to help with a routine logo refresh project. The agreed budget was J$840,000.
The nephew did not assign the brief to Marcia's team. He spent a weekend in Midjourney, generating around three hundred logo variations. He picked twenty, put them through a vector-conversion tool, and presented the cleaned-up final five to the marketing director on Monday. The director picked one. The project was closed in eight days.
Marcia was given the news at a Friday afternoon meeting. The bank was “exploring efficiencies.” Her team's annual budget was being cut by sixty per cent.
The lesson. The work that was your bread and butter is now being assigned to whoever has the cheapest access to the tools. Defending your role means moving up the value chain into the work the nephew cannot do: strategy, brand systems and client psychology. The execution layer is being compressed much faster than the thinking layer, for now.
Lesson 4.2 · Reality Two · The Half-Way Tree Photographer
A photographer in Half-Way Tree, Tamika, shot a magazine cover featuring an established Jamaican author. The author had freckles and a strong jawline and a particular way of holding her face that Tamika spent the morning capturing.
The magazine's art director, working from home, put the cover shot through a generative editor to “clean it up before print.” The freckles were removed. The jawline was softened. The author's particular face, which Tamika had spent three hours bringing to the foreground, was replaced with a generic, smoother version of itself.
Tamika found out when the magazine landed on shelves. She had not been consulted. There was no clause in her contract about post-production AI editing. The author was furious. The magazine apologised privately. The photographer's name was on the cover credit.
The lesson. Your work no longer ends when you deliver the file. Your contracts now need to address this explicitly: no AI-based modification of the delivered files without written approval from the credited photographer.
Lesson 4.3 · Reality Three · The St Andrew Voice Actor
A voice actor in St Andrew, Andre, spent fifteen years building a career on commercial work. He had a YouTube channel where he posted his old reels, behind-the-scenes recording sessions, and the occasional voice-coaching tutorial. Total content: about twenty hours of him speaking in his recognisable register.
A brand he had never heard of launched a campaign in late 2025 using what was unmistakably his voice. A friend sent him the advert. He had not licensed his voice. He had not been contacted.
He confronted the brand. They claimed the voice came from an “AI provider.” The provider claimed the voice was generated from “publicly available training data.”
The lesson. Anything you put on the public internet can be used as training data, with or without your permission. The remedies are weak, so prevention means getting ahead of the cloning by building your own AI version on your terms and licensing it through a platform you control.
Lesson 4.4 · Reality Four · The Liguanea Motion Designer
A motion designer in Liguanea, Carlene, had a steady book of work on Upwork. Her speciality was explainer animations for software companies. Average rate: US$1,800 per ninety-second video.
In mid-2025 she lost her three biggest recurring clients within six weeks. All three had moved to a 22-year-old in Mandeville who was charging US$300 per video, using Runway and a chain of generative tools.
The Mandeville designer's work was not as good. But “not as good” had stopped mattering for the price-sensitive end of the market.
Carlene's options were to drop her rates, lose more clients, or move up the value chain. She chose the third path. She is now charging US$8,000 per video for projects where the brand cares about the craft, and her client list is half what it was, but her revenue is up.
The lesson. The middle of the commercial creative market is collapsing. The bottom is being eaten by AI-augmented operators charging unsustainably low rates. The top, where craft and authorship matter, is widening, which leaves no safe place in the middle.
Lesson 4.5 · Reality Five · The Kingston Producer and the Painter in Mona
The Kingston producer. A producer based in Kingston pitched a six-episode series concept to a streaming platform. The pitch deck included character sketches, a pilot script, and a treatment. The platform passed on the project after three months of meetings.
Six months later, a near-identical concept appeared in development at the same platform with a different writer attached. The producer cannot prove the platform used his pitch as input to brief their internal AI tools. He also cannot prove they did not. The contracts he signed during the pitch process did not address training-data use; the ones he signs now do.
The painter in Mona. A painter who studied at Edna Manley and works from a studio in Mona had built a recognisable style over fifteen years. In 2025, she was scrolling Etsy and noticed prints in unmistakably her style being sold by an account based in Vietnam. The seller had fine-tuned a generative model on her Instagram posts (about four hundred images) and was producing variations on her work for US$15 a print.
She filed a takedown. Etsy honoured it. The seller opened a new account under a different name within forty-eight hours. The painter is still playing whack-a-mole.
The lessons. For the producer: every pitch document now needs an AI clause limiting training use. For the painter: anything posted publicly is training data. Pre-emptive defence (Glaze, Nightshade, watermarking, brand recognition) matters more than reactive enforcement.
Lesson 4.6 · Patterns Across the Six
Pattern one. The harm rarely comes from the AI itself. It comes from a human in the chain (the nephew, the art director, the platform, the cloner) using AI to do something to your work or your career.
Pattern two. The damage is usually invisible until after it is done. None of them had real-time visibility.
Pattern three. The contracts in place pre-2024 do not address any of this. Updating your standard contract is the single highest-impact defence move.
Pattern four. Visibility helps. The painter in Mona is harder to imitate when she is famous, because the imitation will be obvious. Building brand recognition is now legal protection.
Pattern five. The artists who emerge stronger are the ones who moved upstream into work the tools cannot replicate, rather than down-market into price competition.