AI needs to be smaller, reduce energy footprint: Study
Paris (AFP)The potential of artificial intelligence is immense -- but its equally vast energy consumption needs curbing, with asking shorter questions one way to achieve, said a UNESCO study unveiled Tuesday.A combination of shorter queries and using more specific models could cut AI energy consumption by up to 90% without sacrificing performance, said UNESCO in a report published to mark the AI for Good global summit in Geneva.OpenAI CEO Sam Altman recently revealed that each request sent to its popular generative AI app ChatGPT consumes on average 0.34 Wh of electricity, which is between 10 and 70 times a Google search.With ChatGPT receiving around a billion requests per day that amounts to 310 GWh annually, equivalent to the annual electricity consumption of three million people in Ethiopia, for example,Moreover, UNESCO calculated that AI energy demand is doubling every 100 days as generative AI tools become embedded in everyday life."The exponential growth in computational power needed to run these models is placing increasing strain on global energy systems, water resources, and critical minerals, raising concerns about environmental sustainability, equitable access, and competition over limited resources," the UNESCO report warned.However, it was able to achieve a nearly 90% reduction in electricity usage by reducing the length of its query, or prompt, as well as by using a smaller AI, without a drop in performance.Many AI models like ChatGPT are general-purpose models designed to respond on a wide variety of topics, meaning that it must sift through an immense volume of information to formulate and evaluate responses.The use of smaller, specialised AI models offers major reductions in electricity needed to produce a response.So did cutting the cutting prompts from 300 to 150 words.Being already aware of the energy issue, tech giants all now offer miniature versions with fewer parameters of their respective large language models.
For example, Google sells Gemma, Microsoft has Phi-3, and OpenAI has GPT-4o mini. French AI companies have done likewise, for instance, Mistral AI has introduced its model Ministral.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
The National
an hour ago
- The National
Former Builder.ai CEO Sachin Dev Duggal withdrawn from tech conference amid controversy
Sachin Dev Duggal, founder of the now-defunct artificial intelligence start-up has been pulled from the speakers line-up at a technology conference in Paris. Former employees emailed the organisers of the Raise Summit − which ended on Wednesday − with complaints over the nature of the company's demise. In May, Mr Duggal's former company, which had offices in London, New York and Dubai, began insolvency proceedings. An email provided to The National by Raise organisers said Mr Duggal had been invited to speak to hopefully provide an insight into how the company's fortunes turned. But they said they received 'numerous emails", including from employees who lost their jobs, Raise said. 'However, given the security concerns and the fact that we did not have full clarity on the story, we ultimately decided not to move forward featuring him at the event.' This is the latest in a series of incidents that have marked the fall of a company that once achieved unicorn status − receiving a valuation of more than $1 billion − and a significant investment from US technology giant Microsoft. Some workers at the company have told The National that on legal advice, they had been told to preserve documents and other evidence that may relate to a possible criminal investigation. Some also said they had been issued subpoenas from the US Attorney's Office in Manhattan. In 2024, Mr Duggal briefly spoke to The National as he appeared at an event hosted by the Dubai Multi Commodities Centre, which had just announced a new AI centre. 'With this centre, we're helping to unveil a portal to global commerce,' Mr Duggal said then, reflecting on why chose DMCC as an ecosystem partner. 'We didn't want to be a software company just surrounded by other software companies, we wanted to be surrounded by potential customers and peers.' Yet ultimately, failed to meet revenue expectations. The company had promised to make mobile app development as 'easy as buying a pizza' through the use of heavily promoted AI technology, but it struggled to create reliable products, disappointing investors such as Microsoft and sovereign wealth funds in Qatar. To try to stabilise the situation, Mr Duggal requested more investment funds, and an emergency loan, months before he was stripped of the chief executive title. There have been allegations of financial irregularities, while the legitimacy of the company's AI technology has also been questioned in recent months by several news outlets. In May, Mr Duggal posted a message on LinkedIn within a alumni group created on the platform. 'I don't think the story is done yet, but irrespective, if I can be of any help to any of you as you make this transition in life please know I'm only a message away,' he posted in part. 'And more importantly – if I didn't listen, if I was short, if I was unreasonable – I'm sorry.' On main LinkedIn page, comments from unhappy former customers are plentiful. 'What are you doing to get all of your customers their code for completed apps they paid for?' one person wrote. 'Don't believe these fancy conferences, flying around, inviting celebrities … It's all hype and nonsense,' wrote another, referring to the various events where Mr Duggal often dispensed start-up wisdom. A WhatsApp group created by former employees is full of criticisms as well, with some making allegations about not being paid for work done.
Al Etihad
2 hours ago
- Al Etihad
Robot performs realistic surgery 'with 100% accuracy'
10 July 2025 00:10 NEW YORK (dpa)A robot has performed realistic surgery on its own with 100% accuracy, researchers have said. In a "major leap" towards using more robots in operating theatres, a machine trained on the videos of surgeries was able to precisely work on removing a robot operated with the expertise of a skilled human surgeon, according to Johns Hopkins University researchers in the US, even during unexpected scenarios typical in real-life medical robot was watched as it performed a lengthy phase of a gallbladder removal on a life-like was able to respond to and learn from voice commands from the team, just like a novice surgeon working with a there were 17 tasks in the surgery, the robot had to identify certain ducts and arteries and grab them precisely, strategically place clips, and sever parts with scissors. It was also able to adapt even when dye was introduced which changedthe appearance of the organs and professor in mechanical engineering, Axel Krieger, said: "This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures."This is a critical distinction that brings us significantly closer to clinically viable autonomous surgical systems that can work in the messy, unpredictable reality of actual patient care."The work received US government funding and was published in the journal Science Robotics. Back in 2022, a robot performed the first autonomous robotic surgery on a it required specially marked tissue, operated in a highly controlled environment, and followed a rigid, pre-determined surgical new system, which uses the same machine learning architecture that powers ChatGPT, also adapts to a patient's anatomical features in real-time and works to correct can respond to spoken commands such as "grab the gallbladder head" or "move the left arm a bit to the left," and then learns from that the robot took longer to perform the work than a human surgeon, the results were comparable to an expert surgeon, researchers said. Next, the team will train and test the system on more types of surgeries.
TECHx
4 hours ago
- TECHx
High-Severity Windows Flaw Revealed by Security Expert
Home » Emerging technologies » Cyber Security » High-Severity Windows Flaw Revealed by Security Expert Positive Technologies has revealed a high-severity vulnerability affecting 37 desktop and server Windows operating systems. The flaw impacts Windows 11, Windows 10, Server 2025, Server 2022, and Server 2019 across various versions and architectures. The vulnerability, identified as CVE-2025-49689, was discovered by Sergey Tarasov, Specialist at the Positive Technologies Expert Security Center. It affects the NTFS file system driver and was assigned a CVSS 3.1 severity score of 7.8. The flaw could have enabled privilege escalation if a user opened a malicious virtual hard disk (VHD). This would allow attackers to bypass Windows security and gain full control of the system. Microsoft was notified under responsible disclosure protocols and released patches in July 2025. • Over 1.5 million devices are reportedly exposed • U.S. and China account for the largest number of affected systems StatCounter data shows Windows 11's market share rose from below 30% in 2024 to over 43% by May 2025. Tarasov explained that attackers often use VHD files in phishing campaigns. Many users treat these files like ordinary archives, increasing the risk of exploitation. Positive Technologies recommends users install the latest Windows updates. If updates cannot be applied, users should only open VHD files from trusted sources. The company also advises deploying its tools, including MaxPatrol VM and MaxPatrol EDR, to detect and prevent similar threats. In 2024, Tarasov helped address another vulnerability, CVE-2024-43629, affecting Windows 10, 11, and Server editions. In 2017, the PT Expert Security Center collaborated with Microsoft to resolve CVE-2017-0263.
