Jack Dorsey thinks AI-driven personalization could be the solution to Twitter’s problems.
Speaking to investors at the Goldman Sachs Technology and Internet conference in San Francisco, Twitter’s top executive said the embattled company is increasingly turning to AI and machine learning as a way to tailor Twitter to its users’ interests — and hinted that there are much bigger changes to come.
While Twitter has not previously talked much about artificial intelligence, Dorsey said a new emphasis on machine learning and AI internally has helped make the service easier to understand and more relevant to its users.
Twitter, he said, is “still the best place to get a sense of what’s going on in the world … but as we get better and better at applying machine learning and AI to our system, which we really haven’t done in the past — it’s been very mechanical — it gets easier and easier, and we believe that topics and interests are a big part of that future to make it easier.”
Dorsey pointed to the app’s recently added “explore” tab as one area where users can already see the start of more personalized experience tailored around specific topics. “As you exhaust your timeline you can go to explore and you can actually see topics. And the goal there is eventually to become a lot more topic-based and interest forward.”
As for what a more AI-driven approach means for the timeline, Dorsey was less clear, though he hinted that there is much more to come in the year ahead in the way of more personalized experiences.
“I think for the majority of our life we’ve been fairly mechanical and that manifests in making people do the work and now we have a much stronger machine learning artificial intelligence discipline within the company where we can do things for people. We can be more predictive and it allows us to create more experiences; a few are magical and feel like ‘Oh, Twitter is showing me something I wasn’t expecting and it’s great.'”
Twitter, of course, has been experimenting with variations of this approach for some time. The service made the controversial switch to a more algorithmically driven timeline last year — a change the company is now “seeing a lot of benefit from,” according to Dorsey.
More recently, the platform has also adjusted search results to emphasize relevancy over recency and Twitter is also testing a new “top videos for you” section of its timeline that surfaces content from accounts you don’t follow.
If all that’s beginning to sound more like Facebook than Twitter, it’s definitely not all in your head. Figuring out how to make the firehose of news, updates and other Twitter content more manageable for users has long been one of the service’s biggest problems. Adding more machine learning and personalization (a la Facebook) into the mix is an easy way to address those issues, even if the result may upset Twitter purists.
But before you start dusting off #ripTwitter, Dorsey had a few words of reassurance that Twitter wouldn’t “over-AI it” when it comes to a more curated experience.