Beyond AI in governments: hybrid AI applications for the public sector
Does hybrid AI truly possess human-intelligence to solve challenging problems in the public sector? Here are three use cases for digital governments to automate cognitive tasks by making sense of human-readable data.
Reading, thinking, reasoning—these capabilities of the industry’s new “hybrid AI” extend the frontiers of the technology. This human-like intelligence of hybrid AI that blends the statistical and symbolic classes of AI, as shared in an earlier related article, produces a winning formula that helps to solve complex, and non-standardised problems. But amidst all problem owners across various industries, public sectors are perhaps the most challenged.
Multiple stakeholders make public sector innovation tricky
Digital innovation for governments is not an easy feat. As the principal authority of the state, they are accountable not to one group of “customers” but citizens, businesses, and public officers. Governments are constantly challenged to surpass the ever-elevating expectations of citizens who are primed by services rendered by private enterprises; governments are responsible for standards setting and building a reliable and secure tech infrastructure for businesses to flourish; fulfilling the digital government vision also necessitates a digital enabled workforce to supercharge productivity and decision making.
Can governments address all their digital innovation objectives?
The pressures of these challenges have seen governments roll out new initiatives prioritising AI. For Singapore, these aspirations are materialised in a Digital Government Blueprint (DGB) that describes a digitalisation journey towards being an entity that embraces innovation, operates digitally, and fully leverages data to enhance decision-making. Two principles guide this journey—to be digital to the core, and to serve with the heart—alongside six objectives and strategies for three targeted stakeholders. This process towards emerging as a digitalised, Smart Nation is deeply rooted in AI, with all ministries having submitted plans to adopt AI. But this raises the question of what kinds of AI can tackle the DGB’s holistic yet ambitious scope.
The good news is that hybrid AI has proven results in unlocking better for government agencies through the intelligent Omnitive platform. Similarly, this disruptive technology’s capabilities are perfectly in line with the two overarching objectives from the DGB’s 6 explicated outcomes and strategies:
- to transact with citizens and businesses easily, seamlessly and securely;
- to enable public officers to execute high quality decisions and processes in a timely manner, supported by data and automation.
Recap: TAIGER’s two hybrid AI use cases for governments
In the TAIGER x IMDA Digital Concept Series event held this June, TAIGER shared two use cases in the government sector as part of its effort to collaboratively advance the DGB ambition. These use cases stem from the company’s broad experiences from working with almost every government agency in Singapore; this has culminated into a portfolio of customised solutions taking into consideration the public sector’s objectives, operational framework, security compliance standards, and technical infrastructure.
Use case 1: HR recruitment using semantic search and information extraction
Navigating the job market and then manually processing each application are highly cognitive tasks for both prospective candidates and employers—a golden opportunity for hybrid AI.
The automated process first begins with a candidate’s search query on the employer’s website. With Omnitive Search powering the search engine, candidates can enter queries in natural language, e.g. “Are there internships for biz ops in summer?”. Omnitive Search understands the candidate’s intent and context and pulls out the relevant search result by similarly understanding the company’s content repository. It can even offer recommended resources based on its understanding of the candidate, such as a policy document that explains how interns can qualify for permanent contracts after their stint.
Once the candidate moves on to the application with a resume upload, Omnitive Extract comes into play. Even though resumes are one of the most unstructured documents enterprises process, the document processing engine can automatically clean, read, and extract the necessary data-points from the resume. This includes data-points like contact details, academic degree conferred, work experience and period, and even candidate skill sets. Omnitive Extract performs these tasks with greater than 80% accuracy and further verifies any low-confidence extractions with the applicant for maximum accuracy.
Upon receiving the applications, HR teams no longer need to manually parse each resume—costing time and human error—but can quickly and accurately assess the candidates with one structured glance.
Omnitive Extract pulls the relevant data-points from the resume in a structured format, and for further cross-checking and validation.
Use case 2: Intranet-based virtual assistant
The next use case tackles organisational data silos by synergising all data under a unified knowledge management system, accessed by users via a multi-tenancy chatbot. The solution allows employees to immediately resolve queries instead of tediously hunting for hidden information or spending time consulting colleagues. Omnitive Converse’s use of hybrid AI and NLP techniques easily understand “text language”—i.e. slang, typos, abbreviations and variations—so employees feel like they are speaking with a human.
At the back-end, administrators can train Omnitive Converse to learn more types of intents, and even create actionable insights from detailed usage metrics. Omnitive Converse can similarly be deployed to communicate with citizens and businesses with a great degree of flexibility.
Omnitive Converse as a multi-tenant intranet chatbot for employee self-service.
“Better” unlocked by National Heritage Board: AI-powered art discovery and data-management
One government agency that has unlocked “better” with hybrid AI is National Heritage Board (NHB), the custodian of heritage in Singapore. Its online heritage resource repository, roots.gov.sg, contains information about Singapore’s national collection, national monuments, historic sites, heritage trails and multimedia resources. But with this ever expanding content bank, NHB shared in the TAIGER x IMDA Digital Concept Series that a revamp was needed to improve the search experience and content management.
The impetus to improve roots.gov.sg’s front-end digital experience and backend data management saw the application of hybrid AI. Because of its ability to read and process unstructured information with human logic, hybrid AI was effective in identifying data-points—such as People, Places, Organisations and Historic Events—to classify each artefact. Together with hybrid AI solution provider, TAIGER, NHB trained the AI system with over 6,000 records and controlled vocabularies to successfully tag more than 100,000 resources. This tagging process was shortened from a year to one week to with the help of AI.
“We aimed to use (the project) to improve a work process. When we first tried to tag data manually, we realised that it was untenable. So AI, in terms of named entity recognition, helped us to expedite the process very much. If we were to do it, it would probably take a year—versus having AI do it in less than a week”, shared Shaun Wong, in the TAIGER x IMDA Digital Concept Series event.
Within each artefact, Omnitive would automatically identify entities such as Organisation, Person and GeoBuildings, and generate meta-tags to label each record for accurate search.
Hybrid AI’s rapid content management further contributed to new search and recommendation capabilities to enhance content discovery. Auto-tagged resources, advanced search filters and multiple sorting options created a highly precise search experience, perfect for researchers conducting in-depth research. For casual art explorers, the search engine’s ability to understand natural language queries facilitated serendipitous content discovery for a curated user journey.
One of roots.gov.sg’s new hybrid AI features, content recommendation, even enabled cross-promotion of arts and cultural resources. It allowed users to seamlessly gain exposure to more relevant content from partner databases without having to perform a new search. Revamped, the new roots.gov.sg platform saw increases in engagement duration and pages visited, through an elevated showcase of Singapore’s culture.
The next leap towards better growth may be unlocked with hybrid
More and more will be expected of public services in our digital age. In the necessary digitalisation journeys taken by governments, the quality of what technology is adopted, along with why and how it is applied, are of utmost priority. This is why hybrid AI’s ability to solve cognitive tasks from the front to back office with versatility and efficacy will see it become an essential tool for digitising governments.