Nick: Hello Dr. Suman!
Dr. Suman: Hello, Nick.
Nick: Dr. Suman, what’s next in your opinion for public sector in Analytics & AI?
Dr. Suman: Very interesting question, Nick. As you mentioned, the adoption of analytics and AI technologies is definitely going to accelerate this year. I believe public sector organizations will specifically focus on 5 key areas
- Modernize data estate
- Data democratization
- Natural language processing or understanding
- ML Ops or AI Model optimization
- Responsible AI
Nick: Let’s cover each of these areas starting with data estate modernization. What is it and why is it first in your list?
Dr. Suman: To become digital-native enterprise, public sector organizations will need to modernize legacy data systems and migrate to new cloud-based data analytics platforms.
These platforms will incorporate AI-based data engineering, comply with modern data standards like FHIR for supporting interoperability and exchanging data across agencies, driving self-service analytics, and automated predictive modeling.
The modern platform will simplify data estate management, enhance efficiency, and free-up the data trapped in legacy public sector systems.
For example, CDC’s Data Modernization Initiative (DMI) is a multi-year, billion-dollar effort to modernize core data and surveillance infrastructure across the federal and state public health landscape1.
Nick: What about the data marketplace and what will be the benefits of democratizing data?
Dr. Suman: Modernization of data systems will allow public sector organizations to democratize data, where data will be aggregated, managed, shared and analyzed through an online transactional store.
Anyone in the organization will be able to access and use the data that they need to execute their analytics initiatives.
The advances in data engineering, data interoperability, and self-service data management will facilitate cross-agency data sharing in compliance with data privacy rules and policies. This will allow organizations to build more powerful and effective data analytics models.
Nick: Earlier you mentioned natural language processing. How will organizations use it?
Dr. Suman: Several new advancements have happened to natural language processing. To enhance constituent experience and engagement, public sector organizations will start making use of these advancements in the AI space related to complex language and computer vision models.
Generative Pre-trained Transformer 3 or GPT-3 language model has over 175 billion parameters and may become the most common approach for driving these initiatives.
Organizations will use this pre-trained, deep language AI model and explore crowdsourcing data, social media interactions, unstructured scripts etc. to develop AI solutions for public sentiment analysis. They can also develop domain driven virtual assistants and chatbots. They can also extend this to perform fraud and abuse analytics related to UI claims. There are several opportunities for public sector organizations to make use natural language computing models.
Nick: So then MLOPs will enable all the three initiatives that we discussed?
Dr. Suman: That’s correct. MLOps will allow public sector organizations to deliver higher quality AI programs at speed and scale.
Public sector organizations will adopt this practice to bring together data scientists, data analysts, developers and IT operations professionals to build and deploy AI models faster, and with more repeatability and auditability.
Nick: With all this automation, Dr. Suman, how will organizations ensure that their AI is fair and ethical?
Dr. Suman De: This is where Responsible AI comes in. The systemic bias in data and AI models can lead to unintended consequences. Outcomes can be unethical, inequitable, discriminatory or worse.
Responsible AI offers an approach to address and eliminate this bias and will become the overarching framework that will guide an organization’s AI initiatives.
It is about establishing governance that will ensure regular reviews, process transparency, establish lineage and provenance, encourage use of correct tools and balanced data sets are a few areas of focus for Responsible AI.
Nick: So, what is Infosys Public Services doing to help public sector organizations navigate these nexts?
Dr. Suman: We will help our clients navigate these nexts through our automated data-science platform – IHIP.
IHIP offers a unified cloud-first data analytics platform that makes data boundaryless, aligned to new business standards, enables interoperability, and allows organizations to consume and analyze data at speed and scale.
IHIP includes a data marketplace portal . It will enable data scientists, business intelligence and analytics professionals, and everyone in the organization to consume ready-to-query, aggregated, and anonymized data sets quickly.
IHIP has MLOps built on AWS and allows organizations to automate machine learning model development process – right from data acquisition to feature engineering, data wrangling, model building, and deployment.
What is really interesting for me is our work with natural language processing. We are leveraging pre-trained GPT-3/J models and expose them as a service to organizations.
And, we are developing a responsible AI framework as a part of the platform. This will ensure that organizations AI initiatives are explainable and free of bias.
Nick: Thank you Dr. Suman for your insights on what’s next for public sector in AI & analytics and how to navigate it.
Nick: Thank you, our audience, for joining us today. Check out our other what’s next videos on www.infosyspublicservices.com. For additional information, contact us at askus@infosyspublicservices.com
