AI is an environmental disaster in many different ways. One less obvious issue is what Jutta Haider and Malte Rödl term ‘Algorithmically Embodied Emissions‘. This describes the ways that the algorithmic outputs of many everyday AI procedures (search, recommender systems etc) promote and normalize high-carbon practices.

For example, the default assumption made of anyone using the search term ‘trousers’ will most likely be a desire to purchase some trousers. This will result in a barrage of advertising for various new trouser-related products of all shapes and sizes.

The everyday algorithms that we encounter are baked with consumerist ideology and a general tendency toward economic growth and encouraging high-carbon lifestyles. It takes considerable effort to persuade Google to recommend a train journey over travelling the same route by plane or car.

In an immediate sense, algorithms are constantly pushing us toward environmentally-harmful options. In the longer-term they are perpetuating what Haider and Rödl term “search engine-facilitated ignorance (and knowledge) logics related to the climate crisis”. In both ways, we need to encourage a different culture of online information that is more environmentally conscious, if not environmentally-protective.