Dive Brief:

• The enterprise is still far from a technological singularity, but AI can now help create AI. Google introduced AutoML in May, a controller neural net system which proposes and trains machine-generated "child" architectures via deep learning models, according to the Google Research Blog. In other words, a deep learning based machine learning system is being used to create another machine learning system. 
• The AutoML system is reportedly capable of producing systems far more efficient and powerful than those created by human developers, achieving an 82% accuracy rate categorizing image content, reports Wired. Human-developed systems, by contrast, achieved a 39% accuracy rate.
• Google CEO Sundar Pichai noted the lack of machine learning scientists and expertise around the world, and other experts have expressed hopes that automation of some of the work tying up AI experts will free up attention for more complex AI systems, according to Wired. Pichai hopes AutoML will become a democratized tool, opening up AI to more tech workers.

Dive Insight:

Deep learning has yet to hit full stride in the enterprise. It is expected to be an integral part of developer tool kits before 2018, and over the next year it is expected to drive performance, fraud and failure predictions.

An open environment for AI got a huge win Thursday when Microsoft and AWS announced a deep learning partnership and the launch of Gluon, an open source interface to help developers use the technology and ease development of neural networks. The two companies joined Google, Facebook and IBM in the Partnership on AI in 2016, a collaborative research and public education effort for AI.

AI and ML news from tech giants like Google, Microsoft, Oracle and more are frequent, but for the vast majority of companies such advanced tech tools and solutions are not commonplace, let alone possible to develop in-house. Less than 40% of companies have an AI strategy in place, despite majority agreement the tech will confer a competitive advantage.

Pichai's hope for democratized AI tools like AutoML is critical for the expansion of AI around the enterprise. Democratized tools, like Oracle's drag-and-drop AI algorithms, not only open AI up to those with no experience, but also free up developers to focus on other initiatives.