Australian research company DT has developed an on-ear wearable that notifies users when it detects a machine-generated voice, sending a chill down the user’s neck.
As AI becomes more robust, and we’re seeing the implementation of it in various unobtrusive ways that may not be immediately obvious to the human users interacting with it, such as a smart sales assistant that can respond to email enquiries, it is to be expected that methods to detect such AI software would be developed, as we see with an AI-detective from DT’s R&D department.
The Anti AI AI is a wearable that sits behind the user’s ear like a hearing aid. The wearable comprises a small chip from Simblee, an IoT-enabling circuit board designed to make objects easily connectable to smart device apps. The team then developed an iOS app that livestreams the audio (delivered from the wearable via bluetooth) into a machine learning app from TensorFlow, featuring a neural network capable of detecting non-human voices. When a synthetic voice is detected, the wearable sends a signal in the form of a light chill down the user’s neck, delivered via a thermoelectric plate — a tiny refrigerating device that creates a cooling effect by sending a small current across the plate. DT R&D are planning to add in the ability for the neural network to learn from real-world experience, creating an evolving AI synthetic voice-detecting device, but currently have no set release date for their Anti AI AI.
We’ve already seen examples of businesses using AI algorithms that craft communications based on the recipient’s online personality — will there be more counter movements aiming to detect when a company is using these kinds of AI techniques?