Each time you ask Amazon’s Alexa to book a meeting or call mom, your request is sent to Amazon Web Services for processing. Alexa’s intelligence actually lives in Amazon’s cloud.
AI-powered cloud platforms can embed intelligence into services and help businesses benefit from AI’s many applications. Amazon Web Service is the biggest cloud provider, enriched with services such as predictive analytics and fraud detection; and developer tools such “Amazon Lex,” a service for building conversational interfaces.
Oracle is the world’s second-largest software company and hosts the industry’s largest third-party data marketplace. The company, however, has not employed AI in its cloud services to the same degree as industry leaders like Amazon or Google. Oracle Chairman and CTO Larry Ellison are determined to change that by using machine learning.
Oracle’s intelligent database
At Oracle OpenWorld 2017 in San Francisco, the company announced its Oracle AI Platform Cloud Service, which features a suite of Adaptive Intelligent Applications, the Oracle Autonomous Database Cloud, and the Oracle Security and Management Cloud.
The Adaptive Intelligent Apps will drive businesses in finance, human resources and supply chains. By enabling machine learning and data processing capabilities, these apps can learn from data, help businesses to obtain insights from customer data, and adapt in real time.
Oracle’s Autonomous Database Cloud is the first of its kind. While humans stay in the loop by tuning, patching, updating and maintaining other databases, Oracle is using machine learning to teach its database to self-repair and self-upgrade while running, which can dramatically cut costs and reduce downtime to less than 30 minutes a year.
The database cloud will be activated in stages — data warehouse workloads by the end of this year; and capabilities for transactions, mixed workloads, graph analytics, departmental applications, document stores and IoT in the future.
Oracle’s new cloud service also includes an AI-powered security and management function, which can augment clients’ existing security systems by learning for example how to better detect anomalies on a manufacturing line.
How to get your AI? Cloud or on-device?
Huawei recently released Kirin 970, the world’s first AI chip for mobile phones; while Apple unveiled its A11 Bionic chip, which packs a neural engine to run AI applications. With AI right there in your phone, who needs the cloud?
On-device AI is limited by the physical size of chips, and a smartphone equipped with today’s AI-specific chips cannot run complicated applications. Oracle’s cloud platform realizes the potential of running AI on the cloud, where capabilities are not limited by machine size. The cloud can also perform more efficiently in training and inference on machine learning models.
Device-embedded AI chips do however have the advantage of functioning in real-time or non-Internet environments and so we may see a sort of harmony develop between these and cloud AI services, wherein they complement each other.
Is an intelligent cloud the next big thing?
Oracle’s AI Platform Cloud Service may mark the beginning of a new relationship between businesses and the cloud. Through advanced machine learning algorithms, the cloud may be able to study and learn how humans are managing their businesses. It could then conceivably automate all aspects of a business, from design and supply to marketing and sales. Tomorrow’s entrepreneurs may have nothing more to do than brainstorming — the cloud will take care of the rest.
Journalist: Tony Peng | Editor: Michael Sarazen