Bergur Thormundsson
Research expert covering technology in the Nordics, AI and smart cities
Get in touch with us nowAI takes its toll in energy and the environment
AI systems are not easy to build and maintain. They require an enormous amount of energy to run their training algorithms and build their datasets. Some estimates suggest that in training ChatGPT 3.0, one of the most recent and advanced chatbot model, nearly 1300 megawatt hours of energy were required only to train the model. This is a considerable amount of energy spent when one considers the average annual expenditure of energy per capita in Europe, for example. ChatGPT is also expected to have produced some 502 tons of CO2 emissions in 2022 alone, roughly the equivalent of 100 average humans. While not incredible, considering there are eight billion people on the planet, it is still considerable given that ChatGPT is not the only large language model (LLM) in development.It all comes back to sex and gender
In skills and talents, the AI market is still dominated by males. On average men are much likelier to have a tendency for AI skills. It is only in India and the United States that females are more likely to list AI skills than the average individual. This same imbalance is found in the reverse when it comes to the attributes of chatbots, where female-modeled bots are considerably more numerous than male-designed ones. One need not look much further than Amazon’s Alexa or Apple’s Siri to see examples of this. Overall, however, chatbots trend towards a genderless form, with it being the case among nearly 40 percent of chatbots in 2022.
