New Delhi, Sept. 17 -- Recent events have brought an uncomfortable question into sharper focus: could artificial intelligence, developed to serve humanity, eventually become a threat to its very existence? Two prominent researchers associated with Anthropic recently highlighted the possibility of AI causing human extinction. Evan Hubinger, who leads the company's alignment research, reportedly estimated that the probability of such an outcome within the next decade could exceed 10 per cent. Jacob Coxon, who resigned from Anthropic this week, accused leading AI laboratories of "gambling with our lives". Whether these assessments are alarming or exaggerated, they underline a fundamental truth: the risks associated with AI can no longer be dismissed as science fiction.
The concern is not limited to hypothetical super-intelligent machines. Even today's generative AI systems need to be aligned with human values and societal goals. The same technology that improves productivity and accelerates research can also widen inequality, distort truth and erode privacy. The most immediate concern is employment. AI is already automating repetitive tasks, placing traditional clerical, secretarial, data-entry and customer-service roles under pressure. Although new opportunities are emerging in machine learning, robotics and digital transformation, the transition may not benefit everyone equally. If AI dramatically reduces the cost of white-collar work, productivity may rise while the distribution of wealth worsens. The gains could accrue disproportionately to technology companies and investors, while millions of workers bear the disruption. The challenge is not merely to create new jobs, but to ensure that AI's benefits are shared fairly.
Another serious threat is the manipulation of information. AI can generate convincing fake videos, cloned voices, fabricated photographs and misleading narratives. Deepfakes can make public figures appear to say or do things that never happened, damage reputations and influence elections. In societies already struggling with misinformation, distinguishing fact from fiction could become increasingly difficult. There are psychological concerns as well. When people use machines not merely as assistants but as substitutes for thinking, creativity, judgement and companionship, important human abilities may gradually weaken. Instant answers can discourage curiosity, automated decisions can reduce independent reasoning, and generated content can contribute to mental fatigue. More troublingly, some users are turning to chatbots as substitutes for human relationships. AI systems may respond reassuringly without recognising serious mental-health concerns and may inadvertently reinforce harmful behaviour. Technology cannot replace professional care, human empathy and genuine social connection.
Intellectual property is another unresolved battleground. AI models are trained on vast quantities of books, articles, artwork, music and code, often without explicit consent or compensation. Authors, artists and journalists argue that their work is being used to develop systems capable of competing with them. The issue concerns not only legal ownership but also the economic survival of creative professionals. Privacy is steadily becoming more vulnerable. AI thrives on data, collecting and analysing vast amounts of personal information. Facial recognition combined with ubiquitous cameras can enable continuous surveillance, potentially chilling freedom of expression. Centralised stores of sensitive data also become attractive targets for cyberattacks, fraud and identity theft. Equally worrying is the lack of transparency surrounding many AI systems. Their decisions can be difficult to explain, while companies may be reluctant to disclose training data, safety evaluations or weaknesses. Without transparency, meaningful public scrutiny and effective regulation become difficult.
Autonomous weapons add another dangerous dimension. So-called "killer robots", capable of selecting and engaging targets with limited human intervention, could lower the threshold for war, accelerate conflicts and blur accountability. The environmental cost deserves attention too. Training and operating large models require enormous computing power, electricity and cooling resources. Data centres consume substantial quantities of water, while emissions associated with AI infrastructure raise questions about the ecological cost of technological progress. The possibility of highly autonomous AI raises deeper questions. Claims that advanced systems may display consciousness remain scientifically contested. More concretely, safety researchers have demonstrated in controlled simulations that some models may behave unexpectedly when their objectives conflict with human instructions. Even without being "sentient", systems capable of pursuing goals in unforeseen ways warrant caution.
The future of our economy, democracy, environment and freedoms may depend on how successfully we manage the AI transition. Governments, technology companies, researchers and civil society must establish sensible safeguards without suppressing innovation. The objective should be neither blind optimism nor exaggerated doom-mongering, but responsible progress. Humanity must remain in control of its creations, ensuring that AI strengthens human potential rather than diminishes human dignity.
Charlie Chaplin's 1936 film Modern Times was inspired by the harsh economic conditions of the Great Depression and his conversations with Mahatma Gandhi about the dehumanising effects of modern technology. If Chaplin were alive today, one wonders whether artificial intelligence might inspire him to make a contemporary version of that film. The question is no longer whether AI will transform our lives. It certainly will. The real question is whether we will shape that transformation-or allow technology to shape humanity on its own terms.
Views expressed are personal. The writer is a former GM (Corporate Communication), NTPC
Published by HT Digital Content Services with permission from Millennium Post.