The New York Times Features Gado Images
The Times covered our work with OpenAI's Codex and historical archives
I’m thrilled to share that Cade Metz of the New York Times featured Gado Images in a major story about OpenAI’s Codex platform, as well as our 10+ years using Artificial Intelligence to process historical archives.
Read the New York Times story
Codex is a groundbreaking new AI platform that can write computer code automatically using only simple instructions written in plain language. Based on our previous work with OpenAI’s GPT-3, Gado Images was given the exciting job of testing Codex just five days after the system launched.
“As soon as Tom Smith got his hands on Codex — a new artificial intelligence technology that writes its own computer programs — he gave it a job interview,” Metz writes in the Times piece. In Gado Images’ testing, the system performed surprisingly well, writing complex programs--as well as translating between programming languages--automatically.
Want to dig into our testing in more detail? Check out this article I wrote about the technical details of our time spent testing Codex.
We put Codex through its paces in several programming-specific tests. Ultimately, though, systems like Codex are interesting even if you’re not a programmer. Like many of today’s AI tools, it promises to enhance the productivity of skilled humans, rather than threatening them with replacement.
In the New York Times piece, Metz saw a parallel between Codex and Gado Images’ archival AI, which has evolved to help our human researchers rather than replacing their important work.
In Metz’s words, “Gado Images…set out to build a system that could automatically sort through the photo archives of newspapers and libraries…But the technology could handle only part of the job. It could sift through a vast photo archive faster than humans, identifying the kinds of images that might be useful and taking a stab at captions. But finding the best and most important photos and properly tagging them still required a seasoned archivist.”
Ultimately, AI works best when it complements the work of skilled humans--whether those humans are writing code or processing historical photos. Tools like Codex will continue to grow and evolve. Companies that integrate these tools into their workflows in a way that enhances the skills of their talented people will benefit the most from the future of AI.