SHAH ALAM - The emergence of artificial intelligence (AI) in photo editing brings forth both exciting possibilities and ethical dilemmas, according to Universiti Teknologi Mara Intelligent Systems, College of Computing, Informatics and Mathematics (KPPIM), Research Initiative Group, senior lecturer Nurzeatul Hamimah Abdul Hamid.
While AI offers innovative tools for enhancing photos, Nurzeatul cautioned against potential pitfalls akin to those observed with deepfakes.
"Similar to the ethical concerns raised with the introduction of deepfakes, excessive use of AI in picture editing or filters may lead to misinformation or disinformation.
"Limited understanding of how editing works might satisfy our immediate needs, such as appearing younger, but one must understand the consequences as well,” she told Sinar Daily when contacted recently.
She stressed that while a flawlessly AI-edited photo may meet one's desires, it could also result in misunderstanding or provide a false representation.
Nurzeatul emphasised that AI is data-driven, highlighting how the inclusivity and diversity of training data directly influence AI-powered features' outcomes.
She urged AI developers to prioritise these considerations in their design requirements.
In the context of promoting a healthy body image, Nurzeatul pointed out how AI systems like large language model (LLM) or GenAI learn from vast internet data.
However, she cautioned that the concept of a healthy body image may be trained to align with images commonly associated with keywords like ‘healthy body image,’ such as white, slim figures, which may not accurately reflect the diverse definition of a healthy body.
According to Nurzeatul, transparency and inclusivity in algorithm design are crucial.
She advocated for developers to be more transparent and for the development of additional tools that educate users about algorithm workings and necessary precautions.
"To strike a balance between AI tool advancement and ethics is not easy.
"Hence, educating users or the public on how these tools work is crucially important,” she added.
Nurzeatul suggested that developers should be transparent about the foundations on which their applications are built, offering channels for users to provide feedback on app functionality.
In the context of originality, privacy, and consent, Nurzeatul emphasised the need for users to be informed about the original sources of AI-generated outputs.
She also pointed out the importance of initiatives like the AI for Rakyat initiative, aimed at educating citizens about AI functioning and applications.
"Responsible AI involves not just users and the public but also developers, industries, and governments,” she added.
She emphasised the societal impact of unregulated social media and stressed the importance of AI education, especially concerning AI-powered photo and video editing tools.
Nurzeatul cited the advanced frameworks and regulations on AI in the European Union and agencies like Malaysia Digital Economy Corporation (MDEC), and Ministries (Communications and Digital Ministry, MOSTI) are actively working on policies and guidelines.