According to the ‘2025 Top Strategic Technology Trends’ by Gartner, AI risk assessment will be a significant focus area for organizations due to the rise of Agentic AI (a class of AI that focuses on autonomous systems that can make decisions and perform tasks without human intervention), AI governance platforms and disinformation security.
For businesses, this means innovation is required for quicker data analysis and prediction intelligence, assessing privacy violations, guiding and tracking AI models through governance standards and looking out for fraud and phishing.
Now more than ever, businesses need to quickly overcome the challenges of data silos, inconsistent data modalities and integration issues that complicate data analysis and decision-making. Furthermore, the threat of data breaches and cyberattacks creates obstacles to data sharing and collaborative efforts among companies within the same industry.
These challenges are amplified within complex and interconnected digital supply chains. Enterprises are increasingly looking to AI to provide better intelligence that goes beyond incremental productivity gains.
Building better intelligence
So what constitutes better intelligence? Over years of working with enterprise customers, we at Human Managed have learned that intelligence that is particularly effective and trustable has three qualities – it is traceable, timely and fresh. However, to generate such intelligence consistently, a wide range of data needs to be processed, analyzed, applied to well-defined problems and distributed to the right channels in real time.
Enterprises are increasingly looking to AI to provide better intelligence that goes beyond incremental productivity gains.
How can AI lead the way in building better intelligence? Enter Federated Learning (FL) with privacy preservation techniques, a machine learning approach where multiple institutions (for example, banks, hospitals, telcos, insurers) can collaborate within their sector or across industries to train a shared model while keeping their data decentralized and secure. This approach is particularly appealing to industries such as the financial sector, where privacy and security are paramount.
So why would enterprises, especially in the Association of South-East Asian Nations (ASEAN), consider the route to collective intelligence? As our recently launched white paper, titled ‘Better Intelligence Is Collective: Unlocking The Potential Of AI With Federated Learning’ reveals, there are several reasons:
1. The demand for quality data is growing: Reusable, scalable and adaptable AI requires quality data and data privacy is business critical. At the same time, there is also a need to reduce large language model training costs and hence specialized technologies such as FL seem to check all the boxes.
2. Diverse application potential is attractive: A wide array of industries, including banking and financial services, healthcare, mobile applications and autonomous vehicles can benefit from FL and privacy preservation techniques. Successful applications for FL solutions have been used during the COVID-19 pandemic and for financial risk management and manufacturing. Academic research has further highlighted its potential for IoT, telecommunications and healthcare.
3. ASEAN digital economy is poised for growth: While FL is at a nascent stage in ASEAN, the application potential is strong. As one of the fastest-growing regions of the global economy, estimated to reach US$4.5 trillion by 2030, ASEAN’s digital economy could add up to US$2 trillion by 2030.
According to the ASEAN Digital Economy Framework Agreement (DEFA), the interconnectedness of the ASEAN ecosystem can deliver a multiplier effect across many layers. The priorities of the DEFA agreement highlight six significant opportunities: cross-border data flows and data protection; online safety and cybersecurity; digital ID and authentication; digital payments and e-invoicing; cross-border ecommerce; and digital trade.
Challenges ahead
What challenges, incentives and solutions could affect adoption of FL? While tech challenges can be overcome easily enough, participation incentives are key in encouraging enterprises to adopt FL. Additionally, modular technologies will be critical in scaling FL across diverse industries and regions in ASEAN.
1. Data and model convergence are key tech challenges: For the most part, industry players produce heterogeneous data and adopt heterogeneous models within their business context. A FL approach that attempts to converge and agree on standardized feature sets and model architecture could be costly with little buy-in.
2. Centralized agencies are best placed to drive adoption of FL in ASEAN: While the technology exists for FL to be actioned easily, bigger socio-economic challenges lie around participation incentives, centralized stewardship and regulatory know-how. The case for FL is strongest in highly regulated industries such as banking, insurance, telco and healthcare where government or quasi-government agencies can persuade and enable enterprises to work together.
3. Modular tech can help scale collective intelligence across ASEAN: One way of addressing communication, computation and data and model heterogeneity challenges is via modular architectural data platforms. At Human Managed, our Collective Intelligence platform, hm.works, delivers AI-native solutions for cyber, digital and risk problems for enterprises.
This platform is a modular collection of 14 functions and 92 microservices abstracted into infrastructure, software, data and AI stacks. It integrates data from any source and develops AI models for business context and specific use cases. Through FL and AI-powered apps, the HM collective intelligence platform can build a distributed intelligence sharing system for organizations.
Reusable, scalable and adaptable AI requires quality data and data privacy is business critical.
In conclusion, the journey to operationalizing AI is fraught with challenges, primarily revolving around the management and utilization of vast amounts of data. As organizations navigate data silos, inconsistent modalities, and integration issues, the importance of data privacy and protection becomes increasingly paramount.
The rise of FL offers a promising solution, enabling collaborative, privacy-preserving machine learning across industries. This approach not only addresses data privacy concerns but also enhances the quality of data and effectiveness of AI models.
--------
This article was originally published on The CEO Magazine on April 10, 2025