Blame, Shame, and the AI sustainability game
This blog post is in part a response to some of the ecological or environmental criticism circulating around about AI – and, of course, to empty my head of some of the arguments circling around the issue . Recent articles suggest a huge increase in CO2 emissions and water consumption. These worrying reports have prompted the UK government to develop a Digital Sustainability report on the use of energy and water consumption. This has led to findings that that a projected scale of increase for AI technologies would demand fresh water usage of “more than half of the UK’s total water usage” amongst other issues on energy demand too.
The rise of AI seems to be on an unsustainable trajectory also in terms of CO2 production and water use, with a recent report suggesting that in 2025 AI produced as much CO2 as the city of New York. However, I have noticed that methods to mitigate or problemitise these catastrophic effects often apportion blame in the wrong areas and to the wrong people. This is particularly problematic for higher education institutions to have a universal policy on AI usage. This has prompted me to review the context of users/consumers, complicit parties and proponents of AI’s detrimental environmental conditions.
Source: @lindajeun on Threads
I pose a question: Should universities stop using AI altogether?
With recent protests and petitions at many universities, higher education institutions are struggling to produce policies and legislation that will work. Whilst it may seem easy to put a ban of the use of AI at these institutions, there are a number of issues and obligations they have to traverse.
One of the issues is that universities face is the obligation to student’s professional development in a changing world. Imagine you are a course leader or tutor on a Graphic Design or Illustration course. An outright ban would put students at risk of not understanding current tools and methods working with or utilizing AI. Adobe (an industry standard) uses machine image generation and text prompt LLMs in Photoshop, Firefly and Illustrator, whilst Midjourney and Stable diffusion have become popular pipeline process tools for many studios and employers. Would universities have to ban Adobe? Students would be inhibited on learning how current industry professionals are working alongside or around AI generation in the definition of their roles. What things can it do? What things can’t it do that I can? These are important aspects to any professional development and staying ahead in a fast-paced industry. These professional qualifications need to provide ethical and alternative methods to those driven by industry, but equally need to provide knowledge and experience of how to survive in a professional that works in ways that others may find unethical. It’s a line that is difficult walk for everyone but courses, and universities as a whole, should be places to have discussion about the complexities of these opposing forces.
This also, doesn’t cover the use of AI as critique. Some of the most useful critical discourse has come from an in-depth and practice-based demonstration of the flaws, limitations and issues that AI brings up. Universities are also places to discuss and develop critical perspectives. Not all courses are so industry facing, and are build around a need for freedom to discuss ethical, philosophic and creative potentials of technology – perhaps even if the research is critical of the institution or commercial partners. This is an aspect of ‘freedom of speech’ I haven’t seen broached often in the media – the internal critique. Because you are using AI doesn’t mean you can’t be critical of it, and you are not necessary complicity in its success. For instance, if your provocation or critique causes more impact than your financial support, this could be viewed as a net win.
Covert AI use: Another issue is that AI is becoming increasingly concealed. For every marketing demonstration of their product proudly utilising AI, there are also more and more products that conceal their use of AI behind abstractions and obfuscating language. This is rapidly shifting as more companies look to benefit from the AI boom – shifting functionality or re-dressing existing functionality as “AI powered” features. This is largely a marketing shift rather than a technological development but it marks a willingness to demonstrate the sophistication of their product, even if the AI elements are concealed or not entirely visible. https://www.thedrum.com/opinion/2026-is-the-year-ai-goes-invisible-why-the-brands-that-hide-ai-will-win#:~:text=It%20offers%20gentle%20cues%20instead,AI%20do%20the%20work%20backstage.
Undetectability of AI use: Whilst Turnitin now has guidance on the “likely AI-generated” text, the unequivocality of its use isn’t clear – which makes it harder to penalise than for clear plagiarism. Especially now that citations of AI-generated sources (which may be legitimate research or sources otherwise) are increasingly being cited in text-based assessments. Every Google search too now uses AI, and there are a number of common technological tools that also have AI working in the background unknown to it’s users. https://medium.com/write-a-catalyst/ai-in-everyday-life-10-surprising-ways-you-already-use-it-without-knowing-1a1ef6e689d9
IP theft: An issue that is getting a lot of press is the problem of “IP theft” which many high profile musicians and artists are protesting https://www.bbc.co.uk/news/articles/cwyd3r62kp5o. The use of artistic output being used as training data is a form of ‘use’ not encountered before in legal or artistic practice. Is it right if a machine absorbs your work and can perfectly mimick your output as oppose to a human? This is an open question as it feels like plagiarism, or intellectual property theft. Or is it more akin to tribute acts and influence? Perhaps most ominously is the fact that this further devalues the worth of investing in living creatives. Hiring musicians, comedians, artists, dancers, actors, illustrators etc to provide artwork when a machine can do a good enough job quicker and less stress for less money too. The promotion of investing in living artists over dead ones has long been a issue for creatives but it now seems they have machines to deal with… as well as the dead!
With these issues in mind, criticism of AI has largely been targeted at consumers. Citing a complicity or ‘use as endorsement’ arguments raise a few issues with the forced choice of consumer agency. I might cite Rob Jackson’s Bioshock which looks into the ‘forced choice’ rhetoric within computer games and how this is eeking out into media. Technological or consumer agency is low impact for the polluters and technological subjugators like AI companies. Consumers are often presented with a choice of ‘something very harmful to the environment’ or ‘something quite harmful to the environment’. The choice between the two doesn’t bring about ecological or environmental change. For this to happen, people need to protest, petition and get governments and institutions to change laws, procedures and form new international pressure to adhere to environmental standards. This, of course, is harder than blaming another consumer for making the wrong choice. As mentioned above, blame has been directed at universities, who have been slow to react to the ethical and technological conundrum with equivocal stances in many cases.
Perhaps the follow post will be about ecological plans and imaginations (possibly pipe-dreams) for the future of AI infrastructure.
Sources:
https://www.bbc.co.uk/news/articles/ce85wx9jjndo
https://www.theguardian.com/technology/2025/dec/18/2025-ai-boom-huge-co2-emissions-use-water-research-finds
https://sustainableict.blog.gov.uk/2025/09/17/ais-thirst-for-water/
https://www.eesi.org/articles/view/data-centers-and-water-consumption
The rise of AI seems to be on an unsustainable trajectory also in terms of CO2 production and water use, with a recent report suggesting that in 2025 AI produced as much CO2 as the city of New York. However, I have noticed that methods to mitigate or problemitise these catastrophic effects often apportion blame in the wrong areas and to the wrong people. This is particularly problematic for higher education institutions to have a universal policy on AI usage. This has prompted me to review the context of users/consumers, complicit parties and proponents of AI’s detrimental environmental conditions.
Source: @lindajeun on Threads
I pose a question: Should universities stop using AI altogether?
With recent protests and petitions at many universities, higher education institutions are struggling to produce policies and legislation that will work. Whilst it may seem easy to put a ban of the use of AI at these institutions, there are a number of issues and obligations they have to traverse.
One of the issues is that universities face is the obligation to student’s professional development in a changing world. Imagine you are a course leader or tutor on a Graphic Design or Illustration course. An outright ban would put students at risk of not understanding current tools and methods working with or utilizing AI. Adobe (an industry standard) uses machine image generation and text prompt LLMs in Photoshop, Firefly and Illustrator, whilst Midjourney and Stable diffusion have become popular pipeline process tools for many studios and employers. Would universities have to ban Adobe? Students would be inhibited on learning how current industry professionals are working alongside or around AI generation in the definition of their roles. What things can it do? What things can’t it do that I can? These are important aspects to any professional development and staying ahead in a fast-paced industry. These professional qualifications need to provide ethical and alternative methods to those driven by industry, but equally need to provide knowledge and experience of how to survive in a professional that works in ways that others may find unethical. It’s a line that is difficult walk for everyone but courses, and universities as a whole, should be places to have discussion about the complexities of these opposing forces.
This also, doesn’t cover the use of AI as critique. Some of the most useful critical discourse has come from an in-depth and practice-based demonstration of the flaws, limitations and issues that AI brings up. Universities are also places to discuss and develop critical perspectives. Not all courses are so industry facing, and are build around a need for freedom to discuss ethical, philosophic and creative potentials of technology – perhaps even if the research is critical of the institution or commercial partners. This is an aspect of ‘freedom of speech’ I haven’t seen broached often in the media – the internal critique. Because you are using AI doesn’t mean you can’t be critical of it, and you are not necessary complicity in its success. For instance, if your provocation or critique causes more impact than your financial support, this could be viewed as a net win.
Covert AI use: Another issue is that AI is becoming increasingly concealed. For every marketing demonstration of their product proudly utilising AI, there are also more and more products that conceal their use of AI behind abstractions and obfuscating language. This is rapidly shifting as more companies look to benefit from the AI boom – shifting functionality or re-dressing existing functionality as “AI powered” features. This is largely a marketing shift rather than a technological development but it marks a willingness to demonstrate the sophistication of their product, even if the AI elements are concealed or not entirely visible. https://www.thedrum.com/opinion/2026-is-the-year-ai-goes-invisible-why-the-brands-that-hide-ai-will-win#:~:text=It%20offers%20gentle%20cues%20instead,AI%20do%20the%20work%20backstage.
Undetectability of AI use: Whilst Turnitin now has guidance on the “likely AI-generated” text, the unequivocality of its use isn’t clear – which makes it harder to penalise than for clear plagiarism. Especially now that citations of AI-generated sources (which may be legitimate research or sources otherwise) are increasingly being cited in text-based assessments. Every Google search too now uses AI, and there are a number of common technological tools that also have AI working in the background unknown to it’s users. https://medium.com/write-a-catalyst/ai-in-everyday-life-10-surprising-ways-you-already-use-it-without-knowing-1a1ef6e689d9
IP theft: An issue that is getting a lot of press is the problem of “IP theft” which many high profile musicians and artists are protesting https://www.bbc.co.uk/news/articles/cwyd3r62kp5o. The use of artistic output being used as training data is a form of ‘use’ not encountered before in legal or artistic practice. Is it right if a machine absorbs your work and can perfectly mimick your output as oppose to a human? This is an open question as it feels like plagiarism, or intellectual property theft. Or is it more akin to tribute acts and influence? Perhaps most ominously is the fact that this further devalues the worth of investing in living creatives. Hiring musicians, comedians, artists, dancers, actors, illustrators etc to provide artwork when a machine can do a good enough job quicker and less stress for less money too. The promotion of investing in living artists over dead ones has long been a issue for creatives but it now seems they have machines to deal with… as well as the dead!
With these issues in mind, criticism of AI has largely been targeted at consumers. Citing a complicity or ‘use as endorsement’ arguments raise a few issues with the forced choice of consumer agency. I might cite Rob Jackson’s Bioshock which looks into the ‘forced choice’ rhetoric within computer games and how this is eeking out into media. Technological or consumer agency is low impact for the polluters and technological subjugators like AI companies. Consumers are often presented with a choice of ‘something very harmful to the environment’ or ‘something quite harmful to the environment’. The choice between the two doesn’t bring about ecological or environmental change. For this to happen, people need to protest, petition and get governments and institutions to change laws, procedures and form new international pressure to adhere to environmental standards. This, of course, is harder than blaming another consumer for making the wrong choice. As mentioned above, blame has been directed at universities, who have been slow to react to the ethical and technological conundrum with equivocal stances in many cases.
Perhaps the follow post will be about ecological plans and imaginations (possibly pipe-dreams) for the future of AI infrastructure.
Sources:
https://www.bbc.co.uk/news/articles/ce85wx9jjndo
https://www.theguardian.com/technology/2025/dec/18/2025-ai-boom-huge-co2-emissions-use-water-research-finds
https://sustainableict.blog.gov.uk/2025/09/17/ais-thirst-for-water/
https://www.eesi.org/articles/view/data-centers-and-water-consumption
