Super Clouds and AI.
Setting the Context
It is needless to say that AI is one of the most exciting technological paradigms. The investments made in AI/ML have grown exponentially in the last decade. Large enterprises, 85% and upwards, aspire to be data-driven and aim to leverage AI/ML to enable this effectively.
However, AI alone is complicated, process-heavy, and is very hard to get right. How can we make AI more accessible? How can we make AI more general-purpose? This is where Cloud Engineering comes to the rescue.
How does Cloud enable AI?
There are three main advantages that Cloud provides which makes it very compelling to leverage for anything AI/ML.
- Accessibility: Traditionally, implementing AI/ML solutions was were very expensive due to the large volumes of data, software, and hardware requirements of algorithms. With the economics of the Cloud, problems such as image classification, translation, recognition, etc., are now highly accessible for developers to leverage in their solutions/products.
- Processing and Storing Capabilities: Neural network training is 10–20 times faster using GPUs against using CPU. There is built-in agility to Cloud and coupled with its performance at scale it provides the perfect platform to build, test and deploy AI/ML models.
- Availability of Data, a.k.a “Big Data”: With the adoption of cloud-first architecture by many enterprises, the data is not siloed anymore. It is easier to aggregate all of the data effectively to understand a given problem more holistically. For instance, in e-commerce, it is possible to merge identity and behavioral data, online and offline patterns to craft a more personalized experience.
With the rapid progress within AI/ML, the Super Clouds (AWS, Azure, and GCP) now offer specific Cloud Machine Learning Platforms to facilitate the creation of Machine Learning Models. Also, they provide many AI Cloud Services, abstracting all the underlying complexities such as Natural Language Processing, Entity Recognition, OCR, Object Detection, etc., as application programming interfaces. It is important to note that the list of AI Cloud Services is growing at a rapid pace. It is this phenomenon that will enable AI to be more general-purpose over some time.
AI offerings from AWS, Azure, and GCP
Here is a compilation of resources provided by the superclouds compiled by Vaibhav Satpathy in his article. Please refer to official documentation from different cloud vendors to get a complete picture of all the services since many new services are released regularly.
As you can see, there is a wide range of services offered by cloud vendors spanning from recommendations, speech-to-text, video intelligence, etc. The list is bound to improve over some time. Application developers now can easily leverage the benefits of AI/ML with these cloud-native services.
Conclusion
Cloud is enabling enterprises to drive rapid innovation by aiding in and abstracting AI/ML. Over some time, Cloud will make AI more general-purpose with Cloud-Native offerings.
As a developer, if you aspire to be a data scientist, understanding one of the core cloud platforms will be a great advantage. Also, as an application developer, you can start reaping the benefits of AI/ML if you equip yourself with the knowledge of one of the Cloud Platforms.
