MIT Develops AI to Update Wikipedia

Keeping encyclopedia entries up to date can be difficult. Still, MIT researchers believe they’ve found a way to make life easier for those engaged with ensuring the world’s most frequented encyclopedia is up to date. The MIT team used an artificial intelligence agent to locate specific information within Wikipedia articles and updated them while keeping the tone and style of the initial material. Based on current methods of updating the tool should make life a lot easier for members of the edit team. One of the most delicate parts of the process is ensuring that the style matches that of the article throughout. The AI engine ensures that the form remains consistent even with the changes.

No Need for Structured Input

One of the benefits that the engine offers to updaters is the way it takes input. Updaters can simply feed the engine data in an unformalized, unstructured manner. The engine will then analyze the data, locate the entry on Wikipedia, and update that information. The engine isn’t the first bot developed to update Wikipedia entries, but it is by far the most sophisticated. Most other bots utilize a rule-based system. The MIT bot leans heavily on reasoning to generate coherent, readable text. Sometimes that reasoning requires the engine to consider conflicting parts of a sentence and resolve it so that it seems like a naturally written sentence.

Automatically Updating Wikipedia?

The MIT team sees this as the first step in a larger project. The team has stated that the bot has been successful in spotting and flagging fake news. Experts believe that if the system were to be extended and taught how to collect and categorize information itself, the functionality could be extended. At that point, the bots would merely canvas the internet and collect relevant data to update Wikipedia entries themselves. This reality is quite a way off, but the current iteration of MIT’s Wikipedia bot is exciting to witness.