Guests include academic researcher Kate Crawford discussing the true cost of AI, reporter Chris Vallance on DeepMind's watermark system for AI-generated images, Mansoor Hamayun on the smart cooking valve in Rwanda, and Fu’ad Lawal on archiving Nigeria's newspapers. Topics include environmental impact of AI, gas canister system for affordable fuel, preserving Nigerian history through digitizing newspapers, and the impact of policy and cultural expression in AI.
AI systems have a significant environmental footprint due to their high energy and water consumption, as well as reliance on data and hidden labor.
DeepMind has developed a watermark tool to identify AI-generated images, addressing concerns of misinformation and protecting the originality of human-created designs.
Deep dives
The Environmental Impact of AI: Energy, Data, and Mineral Extraction
One of the main concerns regarding artificial intelligence is its environmental impact. Large-scale AI systems, such as GPT, consume an enormous amount of energy, making them up to a thousand times more energy-intensive than traditional search engines. Additionally, the conversation with AI systems like GPT requires a significant amount of water, equivalent to pouring a full liter onto the ground. Moreover, AI's reliance on data and hidden labor further adds to its environmental cost. These systems extract large amounts of data from users, employ crowdsourced workers for labeling and cleaning datasets, and require vast electricity and water resources for their operations. As AI technology rapidly becomes a part of everyday life, it is crucial to examine and address its extensive environmental footprint.
DeepMind's Efforts to Detect AI-Generated Images
As AI image generators become more sophisticated, distinguishing between images created by humans and those generated by computers becomes challenging. DeepMind, a Google company, has developed a tool to embed a digital watermark into AI-generated images. Although imperceptible to the human eye, this watermark remains intact irrespective of alterations to the image. By running the image through Google's image generator system, Imogen, it can determine if the image was created by its own AI. This development aims to combat the spread of misinformation and protect the livelihoods of designers and artists, who are concerned about AI using their work to mass-produce similar images. While this watermarking tool is an experimental solution, it highlights ongoing efforts to address the issue of AI-generated images and their potential impact on various industries.
Addressing Fuel Poverty in Rwanda with Smart Cooking Valve
In Rwanda, a company called Beebox has introduced a smart cooking valve to tackle fuel poverty and reduce reliance on charcoal. This innovative device allows people to access affordable and clean cooking fuel by attaching it to a gas canister and using a mobile payment system. By providing bite-sized and affordable gas consumption options, the smart cooking valve offers an alternative to more expensive gas usage. This initiative is crucial as many individuals and families in developing countries rely on charcoal and kerosene, which have detrimental health effects and contribute to deforestation. Beebox's efforts align with the World Bank's program to eliminate charcoal use, promoting cleaner and healthier cooking methods.
Archiving Nigerian Newspapers for Accessible History
Archive Inc., a project in Nigeria, aims to preserve and make accessible Nigerian history through archiving and distributing newspapers from 1960 to 2010. This initiative addresses the lack of collective memory and limited access to historical information in Nigeria. By digitizing and cataloging newspapers, Archive Inc. enables people to explore millions of stories, fostering curiosity and providing a platform for understanding the country's past. Archiving newspapers is seen as a necessity as they serve as a rich narrative of a nation, capturing stories and events that shape cultural, political, and governance decisions. Moreover, the availability of historical materials can prevent the spread of misinformation, offering a reliable source of information for future generations.
This week, the academic Kate Crawford tells us how she travelled the world to find the true cost of AI. Reporter Chris Vallance updates us on a watermark system - developed by Deepmind, Google's AI arm - which aims to show whether an image was generated by a machine or designed by a human. Mansoor Hamayun, Co-Founder and CEO of Bboxx tells us about the company's smart cooking valve, designed to protect lives - and trees - in Rwanda. We speak to Fu’ad Lawal, the founder of Archivi.ng,and archivist Grace Abraham, about why the key to Nigeria's tech future may lie in digitsing newspapers from its past.
(Picture credit: an imagined digital landscape, by Andriy Onufriyenko, for Getty images)
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
