Advertisement
Sections
Microsoft gives Swiftkey a brain: Neural networks to power all your typing needs
Most Android users must be familiar with Swiftkey, that decidedly awesome keyboard with an uncanny ability to predict your next words. That keyboard was good, but it’s latest update makes it even better.

Most Android users must be familiar with Swiftkey, that decidedly awesome keyboard with an uncanny ability to predict your next words. That keyboard was good, but it’s latest update makes it even better.Swifkey’s prediction mechanism depended on the so-called n-gram model, which only looked at the previous two words before coming up with a suggestion. It also learned from your writing style.It was good enough to beat most keyboards on the market, but n-gram doesn’t understand context. Microsoft, who bought Swiftkey a while ago, have now updated the algorithm to support “neural networks.” That’s a fancy way of saying that Swiftkey now understands context.As Microsoft explains, if you’re typing “let’s meet at the… ,” Swiftkey would only look at “at the” and predict “time” or “moment”, or words of that nature.The updated algorithm (the neural network) will look at your sentence as a whole and figure out what you’re trying to say. The result, claim Microsoft, is a prediction mechanism that’s far more reliable than it’s ever been. In the above example, Swiftkey will now predict “hotel,” “restaurant,” "airport," etc.As with all neural networks today, Microsoft trained Swiftkey with “millions of sentences.” The updated Swiftkey app is available on the Google Play store and currently supports UK and US English languages.
First Published:Sep 16, 2016, 17:55:17 IST
Advertisement
Advertisement

Why AI notetakers are raising serious privacy and security concerns
AI notetakers promise effortless meeting summaries, but experts warn they could expose confidential conversations, corporate secrets and personal voiceprints. As businesses increasingly adopt AI-powered meeting assistants, questions over data storage, privacy, consent and legal risks are becoming impossible to ignore
5 min read
China's low-cost AI models are changing the global AI race. Here's why Silicon Valley is worried
As Chinese firms continue to improve performance while keeping prices low, the AI race is no longer just about building the smartest model—it is increasingly becoming a battle over who can deliver the best value
2 min read
China's Kimi K3 challenges US AI leaders with frontier-level performance at lower cost
Chinese artificial intelligence startup Moonshot AI has unveiled its latest open-weight AI model, Kimi K3, with early results suggesting it could compete with some of the world's most advanced AI systems developed by leading US companies
2 min read
How did Instagram run ads promoting child abuse in India?
India has issued a notice to Meta after an investigation alleged that Instagram displayed paid advertisements promoting child sexual abuse material. MeitY ordered Meta to remove such Instagram ads and explain within seven days how they were approved
8 min read
Why has India halted WhatsApp’s username feature before launch?
India has halted WhatsApp’s planned username feature, citing concerns about cybercrime, impersonation, and law enforcement challenges. As Meta races to address security concerns, here’s why MeitY has paused the rollout, what the feature does, and how it could influence privacy and online safety in India
3 min read
Advertisement
Advertisement
