Press enter or click to view image in full size
I started out with Batman hoodies and scented candles. Sometimes, I’d get weirder requests, like ad copy for a paid course on unlocking Chakras. Mostly, it was ecommerce descriptions.
The variety kept the work fresh, and there was plenty of it. For a teen, the side cash was incentive enough.
After stints in copywriting and content management, I decided to tackle programming. I launched a dozen projects before something stuck, and now I’m on my second SaaS, Wordcab.com.
Wordcab uses the latest language technology to generate human-sounding summaries of sales calls and meetings. And through the months building the platform, I discovered something profound.
This is going to be an unorthodox statement, but the group of writers I managed were more vital than any data scientist I could have hired.
Their creativity, understanding of language, and ability to respond well to editor comments helped create a unique product that is used by corporate teams and professionals every day.
Writers are essential to marketing agencies, publishers, entertainment media. But since when was creative writing a core component of a tech product. One that uses state-of-the-art AI technology?
Dataset Writing: An Unlikely Combination
Datasets are a grueling thing. The typical image is of undergrads or (nowadays) entire marketplaces of low-wage workers.
That might be true for labeling objects in photos, but creating datasets for text generation is a completely different beast.
For one, there are no objective metrics for measuring the quality of generated text — only humans can (subjectively) judge the output. And there is little, if any predictability involved.
This causes startups relying on GPT-3 to sometimes land in hot water.
But at Wordcab, it didn’t feel like my team, which at various times consisted of government employees, fashionistas, business analysts, copywriters, and myself— all passionate wordsmiths — was doing data entry.
We felt like we were writing an experimental novel. A futuristic manual for newly minted humanoids. An instruction set for a precocious CPU on the verge of consciousness.
It was a literary challenge, not a scientific one. Sure, there were strict demarcations, but the process was uniquely suited for a writer. And only a writer.
Nearly every other component of the company — from transcription to hosting the AI model online — has been commodified. Hiring more developers or data scientists wouldn’t have changed much.
A Lucrative Future
While Wordcab is focused on English meetings and sales calls, we will hone in on specific verticals this year. To do that, we’ll need more writers.
We’ll need medical writers to help us create datasets for automatically summarizing doctor dictations, legal writers for summarizing video calls with lawyers, and translators for summarizing the rest of the world.
Summarization saves time. Few professionals have the hours needed to check back on their team’s communications, and a process that can eliminate wading through transcripts or listening to recordings will be invaluable.
And that presents an opportunity for skilled creative writers, to help create machines that summarize the world’s communications. It’s an unexpected role, it pays well, and it’s a challenge.
Writers will be part of our AI future.