Google’s AI is Creating AI – and it’s Better than Company Engineers at it

Artificial intelligence (AI) has already managed to get a few amazing achievements. At its I/O 2017 conference, Google’s CEO Sundar Pichai announced a new project that can be seen as “AI inception”: auto-machine learning, or AutoML. Essentially, AI systems can now create better AI systems.

AI can now create AI

Deep learning techniques, that allow AI to keep learning, involve passing data through complicated layers of neural networks. AutoML is Google’s way of automating the time-consuming process by allowing AI to create these layers.

In a blog post, the company stated:

“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. 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 (…)”

So far, AutoML has only been used to create neural networks for speech and image recognition tasks. When it came to image recognition, AutoML matched Google’s experts, but managed to design better architectures than them when it came to speech recognition tasks.

According to Singularity Hub, some of AutoML’s choices were seen as unconventional, as researchers had considered them inappropriate for certain tasks. AutoML is still not a very practical system as it uses a lot of resources, but it still being developed.

If successful, the project can then help non-experts create neural networks tailored to their particular needs, allowing AI to develop far beyond the scope of what Google engineers are currently able to do. Moreover, the speed in which AI systems are created can be significantly improved thanks to AutoML.

Per a blog post published by Google’s CEO:

“We hope AutoML will take an ability that a few Ph.D.s have today and will make it possible in three to five years for hundreds of thousands of developers to design new neural nets for their particular needs.”

According to Wired, Facebook also has its own AI creating AI, and it is also dubbed AutoML. It can essentially go through various possibilities in order to find the best solution to a given problem.

AI is going beyond our homes and phones

So far, most AI systems developed are used to recognize our voices, identify objects in photos, and decide what content we are looking for on the web, in order to help us perform daily tasks. That is now changing, as recently, advancements in finance have been achieved with the use of AI.

Given that non-experts can now deploy AI systems, and that AI can, in certain ways, now create AI that can surpasses what can be created by human experts, it looks like it’s only a matter of time until AI can determine its own destiny. Although the creepy scenario is a real possibility, there are still a few loose ends. Some are now trying to give machine learning systems the ability to be curious and creative.

Google has already created an AI kill switch, and as technology evolves, AIs are being discouraged from disabling their own kill switches, ensuring humans will always remain in control.

If you liked this article, follow us on Twitter @themerklenews and make sure to subscribe to our newsletter to receive the latest bitcoin, cryptocurrency, and technology news.