Google brings TensorFlow AI software to iOS

Tech behind AlphaGo arrives for iOS developers

Google's TensorFlow machine learning software has now been updated to run on iOS.

The technology that powers AlphaGo was released to the open source community last November, but has now reached the hands of Apple developers who can build a neural network right into apps.

Uses could include face recognition, such as for security, or realistic conversational skills thanks to TensorFlow's natural language tool called Parsey McParseface.

TensorFlow 0.9 is already being used on a wide variety of projects, from practical apps to artistic endeavours such as this neural network-scored theme to an advert for multivitamins.

Google is seeking to move towards an AI-first world with initiatives such as Google Assistant, the world's first mobile assistant with psychological baggage.

Other extras in the new edition of TensorFlow include support for Python 3.5, GPU processing on Mac OS and some additions for the TensorBoard visualisation suite.

It will take some time to see exactly what iOS developers will do with TensorFlow, but its availability for iOS, coupled with its open source nature, mean that it can be moulded to suit, suggesting that we could see a lot of interesting things running on Apple but based on Google technology.

After thrashing a puny human in a Go tournament earlier this year, AlphaGo is now gunning for the claimed best player in the world, Ke Jie.

Ke isn't thrilled about the idea because he is concerned that AlphaGo will learn too much from the experience.

"I don't want to compete with AlphaGo because, judging from its matches with [champion Go player Lee Sedol], AlphaGo is weaker than me. I don't want AlphaGo to copy my style," he said.