KUALA LUMPUR (Oct 26): The real challenge for researchers to tackle misinformation is actually to get the data, said Associate Professor Ika Idris from Public Policy and Management, Monash University Indonesia during the session on Navigating the Disinformation Landscape Amidst Limited Access to Social Media Data.

Social media platforms’ efforts are in vain, as platforms such as X, formerly known as Twitter, are putting their application programming interface (API) behind a paywall, which is too expensive for researchers to gain access to in their efforts of tackling misinformation. 

Using the Indonesian election as an example, Ika said that social media platforms such as Facebook, WhatsApp and Instagram are being used by politicians in the massive orchestration of political narratives while deploying troops to silence criticism targeted towards the government. 

The need for accessibility to data on social media platforms is crucial, as restricting such data will result in investigating actors and networks spreading misinformation. 

“Building the capacity of fact checkers is not sufficient enough. We need to keep pressuring these platforms to provide us access to these data [to tackle disinformation],” stressed Ika.

It is also crucial to have access to multimodal data to avoid biases in machine learning algorithms to increase precision in decision-making, said Dr Mogana Darshini Ganggayah from the Department of Economics and Business Statistics, School of Business in Monash University Malaysia.

Inclusivity, equitability and transparency need to be embraced in the integration of different sources in machine learning algorithms, she said. 

Big data sources are dominating the data landscape, and are leading to centralised data storage that presents challenges and biases, due to lack of integration between different data sources. 

In the session "Machine Learning Algorithms for a Democratic Data Ecosystem", data diversity in integrated centralised data was discussed with the emphasis on the need for better representation to address the challenges and biases in machine learning algorithms. 

Open source tools and platforms play a crucial role in promoting equitable data practices, said Mogana. 

She suggested that more data must be added from under-represented sources, so that the collaboration in sharing data can produce good quality data. Data augmentation, the artificial manipulation of data sets, can be done to increase new data points different from original data points, with transparency and accountability being considered in the process. 

“Make sure you know what your data is telling you, before you do your [machine learning] analysis,” advised Mogana.

Until and unless laws are enacted to regulate artificial intelligence (AI) and big data, there is definitely a threat to human, added Dr Loganathan Krishnan from the Department of Business Law and Taxation, School of Business in Monash University Malaysia in the session "Legal issues on AI and Big Data: Gaps in the Malaysian framework". 

Data is growing rapidly, with it being widely disseminated in the public. As AI is increasingly integrated into the digital economy, calls for regulatory laws in AI and big data to be set in place by the government are evermore so urgent. 

The Personal Data Protection Act (PDPA) 2010 does not cover state and federal government data, which raises issues regarding the accountability of data storage. Moreover, the PDPA does not apply to AI and big data, which showcases the lack of a legal framework and cybersecurity surrounding data received by AI and big data sources. 

With issues such as the ownership of AI-generated intellectual property and the complexity of the level of consent in AI usage of personal data, Loganathan urged the government to formulate a framework to address AI and big data. 

“Even with that, laws are not the only panacea. There must be thick gloves’ enforcement if there is manipulation,” said Loganathan. 

He stressed the importance of decentralising data, being transparent with it, and increasing the nation’s digital literacy in understanding choices they make through AI. Emulating regulatory laws from Europe, the UK, Singapore, South Korea and Canada is also recommended in forming a framework.