Generative AI to supercharge automation in the public sector

The impact that AI will have in the next decade is mind-boggling. In fact, it could raise global GDP by nearly US$7 trillion in that time. 

Mark Geene, Senior Vice President and General Manager of Product Management at automation software provider UiPath, shared this insight as he quoted a study by Goldman Sachs, which predicted that breakthroughs in generative AI could raise global GDP by 7 per cent (equivalent to almost US$7 trillion) over a 10-year period.  

Geene was speaking at the UiPath AI-Powered Automation Summit, which came to Singapore on 27 July. The Summit gathered Singapore automation experts and public sector leaders to share about the vast potential of AI and automation in delivering transformative outcomes. 

Generative AI’s role in revolutionising automation 

At the summit, Geene highlighted how the UiPath Business Automation Platform now incorporates Generative AI to revolutionise automation in public sector agencies.  

UiPath has already helped numerous public sector agencies transform the way they work. For example, Geene shared that UiPath helped the New York state government speed up aid disbursements during the Covid-19 pandemic.  

Automation helped the New York State Department of Labor automate the processing of unemployment aid requests, allowing them to speed through over two million requests in just two weeks. The department was able to eliminate a backlog of 1.3 million claims and helped identify about 800,000 potential fraud cases. 

The integration of Generative AI will allow the platform to deliver even more services to help solve complex business challenges. Geene explained that UiPath has integrated many of the leading large language models (LLMs) available today into its automation platform, including OpenAI’s GPT-4, Google’s PaLM 2 and Falcon AI.  

These models open new possibilities for the automation of different tasks. Take loan processing, for instance. Matthew Tan, Sales Engineering Team Leader, Southeast Asia, UiPath, presented a scenario where a loan officer from Singapore’s Housing Development Board might be tasked with processing a citizen’s eligibility for home loans.  

He explained that Generative AI can help organisations classify the different documents submitted, such as employment records and payslips, and subsequently extract the relevant information needed to assess the applicant’s eligibility.  

Responsible innovation with automation 

A recurring topic during the summit was the need to use AI in a responsible manner. 

“Generative AI is both confidently right and confidently wrong,” Geene said. “These language models are only as good as the data they are fed. Context is what essentially provides these models information to send them in the right direction so they execute a task reliably… without ‘hallucinating’ and making up things that can be damaging to your government.” 

As opposed to LLMs like ChatGPT that draw data from the entire corpus of the internet, public sector agencies may prefer more specialised AI trained on a narrow corpus of data that is verifiable and relevant to the task at hand.  

For instance, a Generative AI software meant to help finance teams will not need to know the recipe for making chilli crab; rather, it just needs to know how to process and manage an invoice, Geene explained. The UiPath Platform allows for organisations to create their own integration if they were to develop their own LLMs to prevent potential misinformation and hallucination. 

The UiPath Platform also ensures that humans are always kept in the loop to ensure accuracy. Going back to the example of the loan officer processing home loans, Tan shared that an individual would still be able to go through all the information extracted to ensure its accuracy.  

Additionally, organisations need to take responsibility for how data is used, said Rob Enslin, Co-CEO of UiPath. A critical consideration is where the data resides. The physical or geographical location of data determines its exposure to diverse laws and threats. Governments may call for data localisation – where citizens’ data is stored within a country’s borders – to address possible privacy issues such as access by foreign governments.  

To allay such concerns, Enslin announced that UiPath now delivers full data residency in Singapore for the UiPath Automation Cloud. This promise of data sovereignty means that sensitive citizen data is now securely protected within the nation’s borders, giving public agencies the reassurance to fully unlock the potential of cloud-based innovation.  

Another important factor to drive innovation is collaboration. Enslin unveiled the UiPath AI-powered Automation GENIUS at the event, a platform that allows public agencies to coach and mentor citizen developers, drive ideation and discovery workshops, and share best practices with one another.  

This will allow public sector leaders to stay updated with the latest innovations and ideate with other like-minded individuals on how automation can unlock exciting possibilities in their organisations.  

Sharing of best practices  

At the event, public sector leaders also had the opportunity to showcase how their organisations have leveraged AI and automation as a catalyst for innovation.  

In the healthcare industry, automation can help doctors provide more personalised care to patients. Wai Ching Chan and Natalia Fernandez Heng, who work on Artificial Intelligence and Automation at the Singapore General Hospital’s  Department of Future Health System,

shared how patient assessment forms facilitated by robotic process automation have improved patient-centric care.

Initially, patients fill out a questionnaire during the appointment, providing key information about their condition. Subsequently, physiotherapists spend up to 10 minutes of a 30-minute appointment slot computing the scores. 

“We transformed these hard copy forms into digital forms – FormSG – and the bot now collates all the responses, performs the necessary calculations, and sends the data to the respective physiotherapists,” shared Chan and Heng. This has saved the hospital 3,700 hours per annum, significantly reducing the time spent on collating patient data and enabling physiotherapists to focus more time on patient care.  

Meanwhile, another case study presented at the event, on citizen service automation, highlighted the use of natural language processing (NLP) to assess public enquiries. An in-house tool harnessed NLP to analyse the text of queries received by a pension and savings agency, and suggested email responses that are reviewed and fine-tuned by the agency’s officers before being sent out. 

Dennis Lui, Chief Executive at VITAL, Singapore’s government central agency for corporate shared services, emphasised the need to view automation as a comprehensive, integrated approach, rather than a quick fix.  

“We need to avoid treating automation as a band-aid solution,” he said. “We need to consider a systems-level approach. That’s where you can unleash the true power of automation. The goal is to maximise the potential of automation and find innovative ways to navigate constraints.”  

The possibilities in the AI and automation journey in the public sector are vast. Yet, with new technology emerging every moment, the runway ahead remains long. Organisations need to be at the forefront of innovation, yet do it in a secure and responsible manner.