One ChatGPT query uses same energy as a second of baking, OpenAI says
SAN FRANCISCO: A single query to the AI software ChatGPT consumes as much electricity as roughly one second of baking something in an oven, according to the developer OpenAI.
Meanwhile the water consumption from each query – owing to data centres needing to be cooled – comes down to about one-fifteenth of a teaspoon, OpenAI chief executive Sam Altman wrote in a blog post.
Experts have for years been warnings about the massively escalating energy demands from the widespread use of AI services. While individual queries may require less energy due to efficiency gains in chip and server technology, the sheer volume of usage continues to drive a sharp increase in energy demand for AI data centres.
Companies such as Microsoft, Google and Amazon are planning to rely on nuclear energy in the US to help meet this demand without proportionally increasing emissions of climate-damaging carbon dioxide.
The need to cool data centres has also raised concerns about water consumption. In recent years, several studies have attempted to calculate the environmental impact of increased AI usage, but researchers must rely on numerous assumptions.
Altman shared OpenAI's energy and water figures in a blog post in which he painted a generally positive picture of the future of AI.
He acknowledged that there would be significant disruptions, such as the elimination of entire job categories. "But on the other hand the world will be getting so much richer so quickly that we'll be able to seriously entertain new policy ideas we never could before."
In recent years, the idea of a universal basic income, funded by productivity gains, has been discussed in connection with AI.
According to the OpenAI chief executive, an average ChatGPT query consumes approximately 0.34 watt-hours of electricity ("about what an oven would use in a little over one second, or a high-efficiency lightbulb would use in a couple of minutes").
He quantified the water consumption at 0.000085 gallons (0.00032176 litres). Altman did not provide further details about the basis for these calculations. – dpa
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
New Straits Times
40 minutes ago
- New Straits Times
Google and Australia's national newswire strike AI deal
SYDNEY: Australia's national newswire has agreed to provide content to Google's Gemini artificial intelligence to improve its responses, the two firms said. Financial details of the agreement, which was announced Tuesday, were not provided. The deal would help deliver real-time information to "enhance" responses in the Gemini app, said Nic Hopkins, Google's head of news partnerships for Australia and New Zealand. Australian Associated Press, which was established 90 years ago, said its journalism would ensure timeliness and accuracy in the information Google's products provide. "This is a strong endorsement of our reputation as a leading and trusted news media organisation," AAP chief executive Emma Cowdroy said. A growing number of deals between news media and generative AI makers have been struck recently as tech firms look to make their natural-language responses to users' questions more relevant. AFP news agency signed a deal with Mistral in mid-January allowing the startup's chatbot to draw on the news agency's articles to formulate responses.
New Straits Times
2 hours ago
- New Straits Times
Ethical fault line: Integrity in the age of AI
WHEN many affirmed it as the next great leap in human innovation, on par with the internet's emergence in the 1990s or television debut in the 1950s. Yet, as emphasised in the recent Awani AI Roundtable, the question is no longer whether AI will transform the media and journalism landscape. The real issue is whether we can control its power without eroding the skills and values that keep society's heartbeat steady. This question arises because AI has already begun reshaping journalism's possibilities. From producing multi-platform content to streamlining newsroom operations, it offers near-limitless capacity to strengthen the profession. The sky's the limit sentiment resonates globally as AI can analyse trends, translate languages in seconds, edit videos and draft articles in moments. Used wisely, it can elevate journalistic thinking, freeing reporters to focus on deep investigations, data analysis and creative storytelling. Yet, this potential comes with a cautionary note: technology is never a neutral force, for it carries the biases, values and blind spots of its creators and controllers. That is why we must be careful. Technology has long been cast as society's saviour, particularly for communities marginalised by geography, economics or politics. AI must remain a tool, not a belief system. Like any tool, AI can be used for either ethical or harmful purposes. In the wrong hands, it risks reinforcing power imbalances, spreading disinformation and establishing media monopolies which are the challenges that require critical human oversight. Oversight matters because the greatest threat may not be spectacular scandals like fake news or deepfakes, but the quiet surrender of our ability to think critically. When AI delivers answers faster than human reasoning, it tempts us to bypass questioning, doubting and innovating. This vulnerability is amplified in education systems that lead compliance over curiosity. Without reform, AI could fill this intellectual space and produce a generation proficient at writing prompts but unable to thoroughly evaluate information. Such risks deepen when AIs development and control rest in the hands of a few global tech giants. This concentration of power risks creating a new form of media colonialism where algorithms shaping our news flows are designed to serve commercial or political interests far beyond our borders. For Malaysia, this is more than an abstract concern. Over-reliance on foreign AI tools, without building local Large Language Models (LLMs), could erode our narrative sovereignty. Journalism must not remain a passive consumer of imported technology; it must become a co-creator, ensuring AI reflects local contexts, languages and values. Without this, our stories risk being reframed to suit someone else's agenda. This concentration of technological power also reshapes the information supply chain, from data collection to content distribution, in ways that are increasingly automated and centralised. Smaller players are pushed aside, while the speed of AI-generated disinformation outpaces human fact-checking. If convincing fake videos can be produced in minutes, we will need verification systems of equal sophistication and transparency so that truth does not perpetually trail behind lies. These risks intersect with another shift: the changing relationship between younger audiences and news. One of the most striking points from the Awani AI Roundtable was that many in Generation Z, despite being digitally savvy, actively practise news avoidance, turning away from traditional reporting altogether. This detachment leaves them more vulnerable to the echo chambers of unimportant content and viral falsehoods. Yet, their deep familiarity with digital tools also positions them to reinvent journalism for the platforms they inhabit. To tap into this potential, we must draw young people into journalism not as content creators chasing virality, but as storytellers serving the public interest. This requires education reforms that integrate AI literacy, ethical reasoning and investigative skills into journalism curricula. AI can accelerate their work, but it cannot replace the human labour of uncovering truths and connecting them to communities. Such preparation is important because the ethical fault line in journalism has never been about speed or efficiency; it is about integrity. Without firm ethical guidelines, AI could normalise corrupt, irrelevant or dangerous practices. The risk of ethical outsourcing emerges when we assume tech companies have embedded fairness and accountability into their systems. They have not. AI mirrors human prejudices and at times, amplifies them. Only attentive human oversight can close these gaps. This raises an urgent question: how should AI be regulated? Should the rules be set globally, nationally or through a blend of both? Global frameworks offer consistency but often fail to account for local realities. National regulations can be more responsive but are vulnerable to political manipulation. A hybrid model, blending international norms with local enforcement and shared accountability between public and private sectors, may be the most practical path. For journalism, regulation must go beyond technical standards. It requires clear rules on transparency, such as labelling AI-generated content; consent, including whose data is used to train AI and correction ensuring harmful outputs can be swiftly removed. This leads to a central principle that AI should not replace journalists but work alongside them. Imagine AI-driven fact-checking tools flagging inaccuracies in real time or algorithms suggesting diverse sources to prevent one-dimensional reporting. These tools can strengthen journalism but only when paired with cultural practices that prioritize human empathy, curiosity and trust; in this way, AI complements rather than replaces journalism. The most effective partnerships occur when AI supports, rather than seizes, the creative process. Newsrooms must be proactive, with clear guidelines covering transparency, accountability and regular bias audits. Journalists should be trained not only in AIs technical uses but also in its ethical and social implications. Crucially, AI can be used to better serve marginalised audiences, delivering tailored content without sacrificing integrity or succumbing to sensationalism. Ultimately, AI will reshape the media industry but the deeper question is whether it will reshape us. Generation Z will inherit leadership roles and their approach to AI will determine journalisms trajectory for decades. They must be equipped to see the forest of social consequences, not just the trees of technological proficiency. In our rush to embrace AI, we must remember what makes journalism irreplaceable: the human ability to ask hard questions, challenge power and connect personal stories to the public good. AI can enhance this mission, but it cannot lead it. As the Awani AI Roundtable concluded, the future of journalism is not an AI-versus-human contest, but a fusion of humans using AI wisely. The sky may indeed be the limit, but our anchor must remain grounded in ethics, community and reality.
Malay Mail
2 hours ago
- Malay Mail
Water and energy are Malaysia's edge in global tech race, says Anwar
KUALA LUMPUR, Aug 20 — Prime Minister Datuk Seri Anwar Ibrahim today said Malaysia's abundance of water and energy resources has made the country an attractive destination for data centre, semiconductor and artificial intelligence (AI) investors. For that reason, Anwar said the water and energy portfolios were placed under Deputy Prime Minister Datuk Seri Fadillah Yusof as they are 'too strategic for the country'. 'If we ask investors, their first two reasons are usually water and energy, and professional workforce is only the third reason. 'We cannot defend our credibility as a peaceful and competent country if we cannot resolve basic issues such as energy and water,' he said at the official launch of the Langat 2 water treatment plant (LRA) at Hulu Langat, Selangor today. The prime minister also stressed the need to upgrade old pipes to reduce non-revenue water and prepare for droughts by exploring modern, water-efficient technologies. Constructed to meet rising demand in the Klang Valley, the Langat 2 water treatment plant can now treat up to 1,130 million litres of water per day (MLD) from Sungai Semantan in Pahang. The raw water is channelled from Sungai Semantan through a 44.6-kilometre tunnel across the Titiwangsa Range to Hulu Langat. The Langat 2 facility will supply treated water to the Western Corridor (KL City Centre, South KL, Petaling, Sungai Besi, Kinrara and Puchong) and the Northern Corridor (Ampang, North KL from Wangsa Maju to Maluri, Keramat and Gombak). The RM4.2 billion project was funded and developed by Pengurusan Aset Air Bhd (PAAB). The plant is equipped with modern technologies including the Granular Activated Carbon double filtration system, a lamella clarifier inclined plate settler, and real-time monitoring through the supervisory control and data acquisition system. It also incorporates environmentally friendly processes such as static mixing, zero discharge, and hosts Malaysia's first and longest pipe conveyor system for water treatment residual. Fadillah, who was also present at the launch, said Malaysia's water supply access stood at 97 per cent in 2024 while the water reserve margin in the peninsula and Labuan was at a satisfactory 14.9 per cent. He added that Malaysia aims to develop local water technologies that meet international standards under the Water Sector Transformation Plan 2040, as well as explore the use of recycled water for non-food industries.
