Introduction
Artificial intelligence has advanced at a breathtaking pace, permeating sectors from finance to healthcare and fundamentally altering the way societies function. Yet behind the veneer of efficiency and progress lies a growing catalogue of harms, including mass surveillance, algorithmic bias, job displacement, and the erosion of human autonomy. This essay argues that, on balance, artificial intelligence will do more harm than good because its risks are systemic, difficult to reverse, and disproportionately borne by the most vulnerable members of society.
AI entrenches and amplifies existing social biases, causing systemic harm to marginalised communities
Explain
Machine learning algorithms are trained on historical data that encodes longstanding prejudices relating to race, gender, and socioeconomic status. When these biased models are deployed in high-stakes domains such as criminal justice, hiring, and credit scoring, they do not merely replicate existing inequalities but entrench them at scale, often with a veneer of objectivity that makes discrimination harder to identify and challenge.
Example
In the United States, the COMPAS recidivism algorithm was found by ProPublica to be nearly twice as likely to falsely label Black defendants as future criminals compared to white defendants, yet courts continued to use it in sentencing decisions. In Singapore, while the government's Model AI Governance Framework encourages fairness, A*STAR researchers have acknowledged that datasets used to train AI systems in Asia often underrepresent minority ethnic groups, raising concerns about biased outcomes in healthcare diagnostics and public service delivery.
Link
The systematic embedding of bias in AI decision-making represents a profound harm that affects millions of people in ways they cannot easily detect or contest, lending strong support to the view that AI will do more harm than good.
AI-driven automation threatens to cause unprecedented levels of unemployment and economic inequality
Explain
Unlike previous waves of automation, which primarily displaced manual labour, AI is capable of performing cognitive tasks, including legal analysis, medical diagnosis, and financial modelling, that were once considered the exclusive domain of highly educated professionals. This means that AI-driven job displacement will be broader and faster than anything the global economy has previously experienced, concentrating wealth among a small technological elite while impoverishing a growing underclass.
Example
A 2023 Goldman Sachs report estimated that generative AI could automate the equivalent of 300 million full-time jobs worldwide, with white-collar professions in law, accounting, and administration among the most vulnerable. In Singapore, despite the SkillsFuture initiative investing over one billion dollars in workforce reskilling, the Infocomm Media Development Authority has acknowledged that mid-career professionals in data entry, customer service, and basic programming face significant displacement risk as AI tools like ChatGPT and Copilot become mainstream.
Link
The scale and speed of AI-driven job displacement, combined with the inadequacy of existing safety nets, strongly suggest that AI will generate more economic harm than benefit for the majority of workers, supporting the proposition that AI will do more harm than good.
AI enables mass surveillance and fundamentally undermines individual privacy and autonomy
Explain
AI-powered surveillance technologies, including facial recognition, predictive policing, and social media monitoring, grant governments and corporations an unprecedented capacity to monitor, profile, and manipulate individuals. This concentration of informational power poses an existential threat to civil liberties and democratic governance, as citizens lose the ability to act, speak, and think freely without the knowledge that they are being watched and assessed.
Example
China's Social Credit System uses AI to aggregate data from surveillance cameras, financial transactions, and social media activity to assign citizens a trustworthiness score, which determines their access to housing, travel, and employment. Even in democratic societies, the use of Clearview AI's facial recognition technology by police forces in the United States, the United Kingdom, and Australia has been criticised for enabling warrantless mass surveillance. In Singapore, the deployment of over 90,000 police cameras island-wide, enhanced by AI-driven video analytics, has raised concerns among civil society groups about the erosion of privacy in the name of public safety.
Link
The capacity of AI to facilitate pervasive surveillance represents a qualitative shift in the power balance between the state and the individual, constituting a harm so fundamental that it outweighs many of the efficiency gains AI provides.
Counter-Argument
Supporters of AI counter that the technology's life-saving applications in healthcare and climate science are so transformative that they dwarf the harms cited by critics. Google DeepMind's AlphaFold solved the protein-folding problem, accelerating drug discovery by years, while AI-optimised energy grids have reduced carbon emissions by up to 40%, suggesting that the net benefit to humanity is overwhelmingly positive.
Rebuttal
However, these benefits accrue primarily to those with access to advanced healthcare systems and clean technology, while the harms of AI, including algorithmic bias, mass surveillance, and job displacement, fall disproportionately on marginalised communities and developing nations. The asymmetry between who benefits and who suffers means that AI's harms are structurally embedded in a way that selective success stories cannot offset.
Conclusion
In the final analysis, the harms posed by artificial intelligence, from entrenching inequality to undermining democratic accountability, are too severe and too systemic to be dismissed as mere growing pains. While AI undoubtedly offers benefits, the asymmetry between who profits and who suffers suggests that, without radical course correction, artificial intelligence will do more harm than good. Societies must therefore approach AI development with far greater caution than they have thus far exhibited.
Introduction
While fears about artificial intelligence dominate public discourse, such anxieties often overlook the extraordinary benefits that AI has already delivered and will continue to deliver across medicine, education, environmental protection, and economic productivity. Every transformative technology in history, from electricity to the internet, has provoked similar alarm, yet the long-term trajectory has consistently been one of net benefit. This essay contends that artificial intelligence will do more good than harm, provided that its development is guided by responsible governance and a commitment to equitable access.
AI has transformative potential to revolutionise healthcare and save millions of lives
Explain
AI's ability to analyse vast datasets, identify patterns, and generate predictions far exceeds human cognitive capacity, making it uniquely suited to medical applications such as early disease detection, drug discovery, and personalised treatment. These capabilities have already begun to translate into tangible improvements in patient outcomes and reductions in healthcare costs, benefits that will only grow as the technology matures.
Example
Google DeepMind's AlphaFold system solved the protein-folding problem in 2020, predicting the three-dimensional structures of over 200 million proteins and accelerating drug discovery research by years. During the COVID-19 pandemic, Moderna used AI to design its mRNA vaccine candidate in just two days. In Singapore, the National University Health System has partnered with AI start-ups to develop algorithms that detect diabetic retinopathy from eye scans with greater accuracy than human ophthalmologists, potentially preventing blindness in thousands of patients annually.
Link
The life-saving potential of AI in healthcare alone constitutes an immense good that is difficult to offset against the harms attributed to the technology, undermining the claim that AI will do more harm than good.
AI drives economic productivity and creates new industries, generating net employment gains over time
Explain
While AI displaces certain jobs, it simultaneously creates entirely new categories of employment and dramatically increases economic productivity, a pattern consistent with every previous technological revolution. Moreover, by automating tedious and repetitive tasks, AI frees human workers to focus on creative, interpersonal, and strategic roles that are more fulfilling and more difficult to automate.
Example
The World Economic Forum's 2023 Future of Jobs Report projected that while AI would displace 85 million jobs by 2025, it would create 97 million new roles in fields such as AI ethics, data analysis, machine learning engineering, and human-AI interaction design. In Singapore, the Smart Nation initiative has generated a thriving ecosystem of AI start-ups and technology firms, with the government reporting that the information and communications sector grew by 7.5 per cent in 2022, creating thousands of high-value jobs that did not exist a decade ago.
Link
Historical evidence and current projections suggest that AI will be a net creator of employment and economic value, challenging the narrative that it will do more harm than good by destroying livelihoods.
AI can be a powerful tool for addressing global challenges including climate change and resource scarcity
Explain
The complexity and scale of challenges such as climate change, food insecurity, and energy management demand analytical capabilities that exceed human capacity. AI systems can optimise energy grids, improve agricultural yields, model climate scenarios with unprecedented accuracy, and accelerate the development of clean technologies, making them indispensable tools in humanity's most urgent existential battles.
Example
Google DeepMind's AI reduced the energy used for cooling its data centres by 40 per cent, a model now being replicated across the technology industry to reduce carbon emissions. In agriculture, AI-powered precision farming platforms developed by companies such as John Deere have reduced pesticide use by up to 90 per cent while maintaining crop yields. Singapore's National Environment Agency has deployed AI systems to predict dengue outbreaks by analysing weather patterns and mosquito breeding data, enabling pre-emptive public health interventions that have measurably reduced infection rates.
Link
AI's capacity to address existential threats like climate change represents a good of the highest order, and it is precisely these large-scale applications that tip the balance in favour of the argument that AI will ultimately do more good than harm.
Counter-Argument
Critics of AI argue that algorithmic bias and mass surveillance represent systemic threats that cannot be regulated away, pointing to the COMPAS recidivism algorithm's documented racial bias and China's AI-powered Social Credit System as evidence that the technology inherently concentrates power and entrenches inequality.
Rebuttal
Yet the existence of bias in AI systems reflects flaws in the data and governance structures, not in the technology itself. Singapore's Model AI Governance Framework and the EU's AI Act demonstrate that robust regulatory frameworks can mitigate these risks, just as safety regulations transformed the aviation and pharmaceutical industries from dangerous to remarkably safe over time.
Conclusion
Ultimately, while artificial intelligence poses genuine risks, the weight of evidence suggests that its potential to alleviate suffering, accelerate scientific discovery, and raise living standards vastly outweighs its dangers. The key lies not in rejecting AI but in governing it wisely, ensuring that its benefits are broadly shared and its harms are actively mitigated. Viewed in this light, artificial intelligence will do far more good than harm for humanity.