Artificial Intelligence (AI) is still young and developing, but that doesn't mean that it can't create another AI.
Out of Google I/O 2017 conference, CEO Sundar Pichai recalled how his team had joked that they have achieved "AI inception".
That is by using 'AutoML' a deep learning project capable of creating smarter AI.
Machine learning involves the technique that passes data through layers of neural networks. Because creating these layers is complicated, Google's initial idea for AutoML was to create AI that could do it for them.
"We are excited about designing better machine learning models, but these days it is really painstaking," said Pichai.
"In our approach (which we call ‘AutoML’), a controller neural net can propose a ‘child’ model architecture, which can then be trained and evaluated for quality on a particular task," explained Google on is Research Blog. "That feedback is then used to inform the controller how to improve its proposals for the next round. We repeat this process thousands of times - generating new architectures, testing them, and giving that feedback to the controller to learn from."
Google has used AutoML to design networks for image and speech recognition tasks.
To see how good it is, the system matched Google’s experts in image recognition. But in speech recognition tasks, AutoML exceeded human experts. AutoML designed better architectures than humans were able to create.
AutoML is Google's step to create AI that is able to help human in developing better machine learning technologies. And according to Google, AI is now at the stage where it's ready to replicate and improve on itself.
"If we succeed, we think this can inspire new types of neural nets and make it possible for non-experts to create neural nets tailored to their particular needs, allowing machine learning to have a greater impact to everyone," said Google.
With its ability, AutoML has the potential to impact other AI and machine-learning software. It could improve speech recognition, facial recognition, image recognition and other Google products.
AutoML is part of a project called Google.ai, which see the potential and gains of AI, and how it can benefit every single product across the Google portfolio. The most important of these, said Pichai, are Google Search and Google Assistant, as well as how the two will work together.
"We are evolving Google Search to be more assistive to our users," he said.