Researchers are struggling to replicate AI studies
Missing code and data are making it difficult to compare machine-learning work—and that may be hurting progress.
The problem: Science reports that from a sample of 400 papers at top AI conferences in recent years, only 6 percent of presenters shared code. Just a third shared data, and a little over half shared summaries of their algorithms, known as pseudocode.
Why it matters: Without access to that information, it’s hard to reproduce a study’s findings. That makes it all but impossible to benchmark newly developed tools against existing ones, so it’s hard for researchers to know which direction to push future research.
How to solve it: Sometimes a lack of sharing may be understandable—say, if intellectual property is owned by a private firm. But there seems to be a wider-spread culture of keeping details under wraps. Some meetings and journals are now encouraging sharing; perhaps more ought to follow.
Deep Dive
Artificial intelligence
Large language models can do jaw-dropping things. But nobody knows exactly why.
And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.
OpenAI teases an amazing new generative video model called Sora
The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.
Google’s Gemini is now in everything. Here’s how you can try it out.
Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.
Google DeepMind’s new generative model makes Super Mario–like games from scratch
Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.
Stay connected
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.