What is AI, how do apps like ChatGPT work and why are there concerns?

Artificial Intelligence (AI) has transitioned from a futuristic concept to an integral part of our daily lives over the past decade. Its applications are vast and varied, ranging from the personalized content streams on social media and the facial recognition that tags our friends in photos to its accelerating role in groundbreaking medical research. However, the recent proliferation of sophisticated chatbots, such as OpenAI’s ChatGPT, has ignited a complex debate, raising significant concerns about the technology’s environmental footprint, ethical implications, and the responsible use of data. Furthermore, alarming claims suggesting that AI could surpass human control and pose an existential threat to humanity have amplified calls for a more cautious development trajectory and stringent regulatory frameworks.

At its core, Artificial Intelligence empowers computers to process immense volumes of data, discern intricate patterns within that data, and execute precise instructions based on the insights derived. It’s crucial to understand that computers, in their current form, do not possess the capacity for genuine thought, empathy, or independent reasoning. Instead, scientists have engineered sophisticated systems designed to perform tasks that traditionally require human intelligence. These systems strive to replicate the human processes of acquiring and applying knowledge. This can manifest in predicting a potential online shopper’s next purchase based on their past buying habits, thereby offering tailored product recommendations. AI is also the silent engine behind ubiquitous voice-controlled virtual assistants like Apple’s Siri and Amazon’s Alexa, and it plays a pivotal role in the development of autonomous driving systems for vehicles. Platforms such as Facebook, TikTok, and X leverage AI to curate and present content to users, while streaming services like Spotify and Deezer employ it to suggest music tailored to individual preferences. The medical field has also significantly benefited, with AI assisting scientists in the early detection of cancers, the review of X-ray results, the acceleration of diagnostic processes, and the identification of novel treatment strategies.

Generative AI, a particularly advanced subset of AI, is capable of creating novel content that can be remarkably indistinguishable from human-made creations. This capability is achieved by training these models on vast datasets of existing information, including extensive collections of online text and images. ChatGPT, alongside its Chinese counterpart DeepSeek’s chatbot, stands as a prominent example of generative AI tools that can produce a wide array of content, including text, images, and computer code. Similarly, Google’s Gemini and Meta AI are adept at engaging in text-based conversations with users. Dedicated applications like Midjourney and Veo 3 specialize in generating images or videos from simple textual descriptions. Generative AI is also making inroads into music creation, with songs mimicking the styles of famous musicians gaining viral traction, sometimes leaving listeners questioning their authenticity.

Despite these impressive advancements, a growing chorus of experts is urging a deliberate shift in focus towards addressing the more immediate and tangible risks associated with AI. These concerns include the technology’s potential to perpetuate and amplify existing societal biases, leading to discrimination against certain demographic groups. This issue stems from the fact that the vast public datasets used to train many AI systems often reflect ingrained societal prejudices, such as sexism and racism. Furthermore, while AI tools are becoming increasingly sophisticated, they remain susceptible to errors. Generative AI systems are particularly known for their propensity to "hallucinate," presenting inaccuracies as factual statements and even fabricating sources for misinformation. This was underscored in early 2025 when Apple was compelled to withdraw an AI feature after it inaccurately summarized news app notifications, and Google faced considerable criticism for the factual errors presented in its AI search overviews.

The integration of AI into educational and professional environments has also sparked debate. Students are increasingly utilizing AI tools to complete assignments, leading to concerns about academic integrity. Conversely, individuals have sometimes faced wrongful accusations of using AI when their work was original, and employees have been found to be "smuggling" AI tools into their workplaces. A significant ethical challenge has also emerged from the creative industries. Writers, musicians, and artists have vocally opposed the unchecked use of AI, citing ethical objections. Numerous creators have issued public statements and signed open letters, accusing AI developers of exploiting their work for training purposes without explicit consent or fair compensation. In October 2024, a significant statement was signed by thousands of creators, including prominent figures like Björn Ulvaeus of Abba, authors Ian Rankin and Joanne Harris, and actress Julianne Moore. This collective voiced strong concerns, labeling AI as a "major, unjust threat" to their livelihoods.

The environmental impact of AI is another pressing concern. While precise figures on AI’s overall energy consumption are still elusive, some researchers project that the industry’s collective energy demand could soon rival that of the Netherlands. The creation of the powerful computer chips essential for running AI programs is a resource-intensive process, demanding significant amounts of energy and water. The escalating demand for generative AI services has, in turn, driven an increase in the number of data centers required to power these operations. These vast facilities, housing thousands of computer servers, often consume substantial energy and necessitate large volumes of water for cooling. While some major technology companies are investing in innovative solutions to reduce water consumption and reuse, or opting for alternative cooling methods like air-cooling, some experts and environmental activists fear that AI’s growth will exacerbate existing water scarcity issues.

The legal landscape governing AI is still evolving, with various governments taking different approaches. The European Union’s Artificial Intelligence Act establishes stringent controls on high-risk AI systems used in critical sectors like education, healthcare, law enforcement, and elections, even outright banning certain AI applications. In China, developers of generative AI are mandated to protect citizens’ data and promote transparency and accuracy, while also adhering to the country’s strict censorship laws. Many nations, including the United Kingdom, have opted to regulate AI through existing legal frameworks, acknowledging the rapid pace of AI development and the potential for new legislation to become quickly outdated. The UK has implemented targeted amendments to existing laws to address specific AI threats, such as prohibiting tools that can generate deepfake nude imagery or child sexual abuse material. Furthermore, the UK has established an independent AI Security Institute tasked with identifying risks and evaluating advanced AI models. Collaborative efforts with the United States are also underway to develop robust AI testing methodologies. However, notably, neither country signed a global AI declaration in early 2025 that advocated for an open, inclusive, and sustainable approach to the technology.