Writing
London Review of Books:
- Horny Robot Baby Voice: On AI Chatbots
- As Long as the King’s Arm? Jiggers, Rods and Barleycorns
- Keep It Clean: The Patron Saint of JPEGs
- Instrumental Tricks: Prosthetic Brainpower
- Its Rolling Furious Eyes: Automata
PORT Magazine
The Guardian
- The DeepSeek panic reveals an AI world ready to blow
- Boris Johnson’s move to bring back imperial units is pure piffle
The Financial Times
- First Person: “Hollywood is turning my blog post into a film”
- Radio Haiti: From evangelists to voodoo priestesses
LitHub
The Verge
I’ve spent most of my writing career (eight years) working at The Verge, where I started as a general news writer before specializing in AI and robotics, becoming a senior reporter and then a senior editor. While at the site I wrote and edited everything from silly blogs (I love blogs) to beat reporting and long-form features, as well as a fair amount of videos and podcasts. You can see my whole back catalogue of some 4,000 stories here or some select higlights below.
When Microsoft released its AI chatbot Bing (with “hidden” personality Sydney) a number of prominent tech journalists were entranced by the possibility that this was sentient AI. In this opinion piece I pour cold water on these romances, arguing that ever since the early days of artificial intelligence, we’ve tricked ourselves into seeing more intelligence in machines than is actually present. It’s a failure of the mirror test — where animals must learn to recognize when they’re looking at themselves.
In 2021, a number of EU parliamentarians reported that they’d been tricked by a deepfake claiming to be Leonid Volkov, chief of staff to then-imprisoned anti-Putin politician Alexei Navalny. However, my reporting showed that this was a case of fear of technology outstripping its actual capabilities, when I tracked down the Russian pranksters who tricked the politicians using nothing more than makeup and dodgy camera angles.
“They call it ‘the hive,’ or ‘the grid.’ Or sometimes just: ‘the machine.’ It’s a huge structure that fills a warehouse on the outskirts of Andover, a small and quiet town in southeast England. It’s impossible to take in at a single glance, but standing on a maintenance walkway near the building’s rafters, you look over what seems to be a huge chessboard, populated entirely by robots.”
One of the most overlooked aspects of the AI boom has been the use and misuse of data. AI researchers often take a cavalier attitude towards this valuable resource, scraping photos, videos, and text from the web without consent. In this investigation I looked into an egregious example: a dataset of faces scraped without permission from transgender YouTubers. These YouTubers shared their transition journeys as a way of connecting with their community, but had no idea their personal content would be used to improve facial recognition systems for use by law enforcement and governments.
During the first COVID lockdown conspiracy theories collided. Across the UK, telecom engineers repairing newly-essential internet cables found themselves confronted and threatened by conspiracy theorists who believed that 5G towers were helping to spread the coronavirus. I spoke to the workers who found themselves on the frontline of an invisible war.
Tales of human-like automata and robots are almost as old as mythology itself. But why are we so entranced by machines that look and act just like us? To find out I took a trip to the home of Engineered Arts — a robotics company in Cornwall in the UK that builds some of the most realistic robots the world has ever seen.
Sometimes, the most important stories aren’t those that break news but add nuance to what’s already been reported. In 2017, at the tip of the AI explosion, a group of scientists claimed they’d invented an “AI gaydar” — able to identify someone’s sexual orientation from a photo of their face. I dug into the work and found that the research was far from as reliable as some outlets had claimed.
Some stories fit into a headline, but it’s the details that make them worth reading. For decades, scientists struggled with Excel and its habit of auto-converting the names of genes into dates. One 2016 study found that a fifth of published papers had been affected by this unintentional error. In August 2020, scientists went so far as to rename a number of genes to stop this problem in future. In the battle of Microsoft versus the Human Genome, it was the geneticists blinked first.
In 2018, the auction house Christie’s sold an unusual portrait of an imaginary nobleman that it claimed was “created by artificial intelligence.” The piece was expected to fetch a modest sum, around $10,000, but fervor for all things AI meant it ended up selling for a staggering $432,500. In the run-up to the auction, however, the AI art world cried foul — revealing that the portrait had been created using code borrowed from another artist. If AI didn’t make the art, which human did?
The arrival of generative AI models that can create images, text, and video jeopardizes a number of professions. But thanks to AI companies’ tendency to scrape up training data and consider the consequences later, it seems copyright lawyers won’t be out of work any time soon. This explainer from 2022 digs into the nitty-gritty of AI copyright law and concludes that nobody knows what will happen next. That’s one prediction about the future that’s generally easy to get right.