AI as an Infrastructure Shared Service
Avi Bender
Executive Advisor for Data Driven Business Process Innovation. Former Director of NTIS, and the first CTO of the US Census Bureau, US Department of Commerce.
Avi Bender
(republished from the ACT-IAC Innovation Newsletter April 2024)
Now that we are all in agreement that AI will transform government operations how do we get started?? Each agency is currently evaluating AI opportunities in cybersecurity, internal business process improvements,? and in meeting their specific mission objectives to better serve the public.? It seems to me that much of the focus is directed at the agency’s application layer and that may not be an optimal approach in the long term.? There are currently data apps and algorithms that can already support applications for fraud detection and other data centric applications across the Federal government. ?Using AI for these bespoke solutions, by well-funded agencies, ?seems like a good start.? But not all agencies have the financial and people resources to build out both the infrastructure and the applications.? Funding constraints and lengthy procurement cycles will hamper an agency’s ability to be nimble and innovative.? The good news is that the commercial marketplace provides a great roadmap for leveraging AI through the data center infrastructure.
领英推荐
I believe we should take a macro view of the civilian government as a “sovereign nation” to create a common (remember shared services) AI infrastructure to support large language models and machine learning.? That infrastructure should be a managed AI data center operation, either on or off premise- for the entire civilian government.? The sovereign knowledge of Federal employees, their contractors, coupled with the power of an AI infrastructure will be powerful and transformative- for all agencies.? Instead of applying AI in discrete use case instances for the benefit of one business unit why not think big and leverage a much broader knowledge set for the whole government.
For many years the value of cross agency shared services was sold as a cost-effective capability that could be shared by many agencies.? ?That business model met limited success with Cyber, ?HR and payroll systems. To further encourage shared services contract vehicles were also established to make procurement a better shared services across agencies.? However, these instruments for acquiring and sharing like procurement services are procedural and legal with a focus on the mechanics of the acquisition and the law.?These essential acquisition instruments typically created by one agency for the benefit of many, may actually fragment both knowledge and attempts at innovation because each agency operates within its own space of knowledge and perceived “unique” requirements.? We have an opportunity to think big and intelligently about business outcomes that leverage new tools in AI – for the whole government- rather than operating within agency moats that were created before AI changed everything.
Avi Bender is the Managing Director of Bender Advisory Services LLC. Avi was formerly the Director of NTIS US Department of Commerce, the first CTO of the US Census Bureau, and the Director for Enterprise Architecture at the IRS.