I am interested in the connection between CloudBank and the National Artificial Intelligence Research Resource (NAIRR) program initiated by the US federal government. What are the pros and cons of using the public cloud (AWS, Azure) for my AI-focused research?
The public cloud is a rich, powerful computing environment that features many AI-specific tools. In the context of NAIRR the initial participating cloud providers are Microsoft Azure and Amazon Web Services. CloudBank works with NSF to facilitate cloud access in the NAIRR program.
There are two qualifiers to be aware of in considering using the cloud. First: Some effort is necessary to understand and manage cloud spend; which works on a utility model of ‘pay for what you use’. Cloud use in the NAIRR program is subsidized by generous credit allotments from the cloud providers; and while this is great news for cloud users, cost management skills are still considered a necessary part of program participation.
Second the skills needed to build research infrastructure also apply to the cloud; but there is an additional necessary ‘skilling up’ process. There is a wealth of documentation, tutorial content and live support available to help a research team develop necessary ‘cloud builder’ skills; but the end result – a working computing environment for research – is ultimately the responsibility of the research team to create and operate.
With these qualifiers in mind we encourage interested PIs to investigate cloud use. Cloud providers are full speed ahead on Artificial Intelligence technology; for example where Azure is integrated with Open AI and AWS has introduced their Bedrock Machine Learning platform that features a variety of AI Foundation Model options. To learn more about the potential of the cloud check the introductory videos found here at the CloudBank portal; and you can also contact our help desk help@cloudbank.org.