This has been the hardest blog in this short series where I have experimented with dialogical research using AI by speaking directly to it. Primarily because I don’t trust it not to contain an implicit agenda one little bit, and as clever as it seems, it can only ever reply algorithmically. So, the answers it gives to my questions are formulaic and don’t say anything that could not be read anywhere on the internet – in effect, my ‘research’ would be little more than the shallow data-mining which plagues and undermines the humanities (Braidotti, 2013). Mind you, many or most people behave algorithmically too, as discussed briefly in the National Portrait Gallery‘s podcasts about AI (2023). I wonder if, in the main, I feel the same about trusting human responses.
Perhaps the difficulty I feel indicates a lack of confidence in these ‘dialogues’. Vilém Flusser, my current favourite writer on technology, tells us all images are dialogical – and what is text on a screen, if not an image? There is much to say about that, of course, but not here, as I try to remain focused on the difficulties of entering into a dialogue with technology about its catastrophic carbon footprint. Except that technologies allow our inner dialogues to exist outside, and always have done – yes, language itself is a technology (Poster in Flusser, 2011; Clarke, 2018.)
While preparing this post and becoming increasingly bored by its replies, I tried asking ChatGPT to write in the style of whomever, but in this instance, the responses came across as satire (try writing something serious and then getting it to translate that into Nietzschien or Pinteresque prose). Satire and climate are a tricky mix, so I have chosen not to reproduce those chats. And in any case, I am not interested in debating the “inviolably human qualities of “real” creativity or […] in nitpick[ing] each new creation’s failures as “proving” that AI will never master “real” art” (Davis, 2022) or anything ‘real’ for that matter.
But there is simply no getting away from the fact that the ecological cost of this latest iteration of technology, the large language model and related functions such as image-generation, is enormous.
The training of complex deep learning models requires a substantial amount of computational power, often leading to a significant carbon and water footprint. And the rapid growth of the AI industry has led to an unsustainable rise in demand for hardware and the raw materials used to make it, which takes a massive toll on the planet’s air and soil quality.
AI Has a Huge Climate Change Problem, Glover E, 2023
There are those who will undoubtedly frown upon me for even engaging with Artificial Intelligence at all, given its obvious climate related costs, not to mention the appalling lack of care for millions of people employed to sanitise data, or the deep concerns about copyright theft, and more besides. However, since AI is increasingly employed across sectors, sometimes in ways that aren’t always obvious, I’d challenge those critics to live without it. I also think the only way to address these concerns is to engage with the technology and to try and understand how and why it’s being invented in the first place, and then used by so many people and across all sectors.
I will report that ChapGPT told me simply (but only after I asked it to use fewer words) to advocate for responsible AI practices, transparency, positive change. And to balance innovation with sustainability and ethics.
Unfortunately, for the naysayers, and perhaps tentatively happily for the rest of us, the sorts of calculations that lead to AI generated images and chat bot conversations are also extremely suited to the discovery of energy efficient solutions required in our modern burning world. The algorithms that can generate patterns potentially also provide complex scientific solutions, as has been seen in the pharmaceutical sector with protein construction (Schwaller, 2023). It could also help generate materials that preserve heat, or servers that can function at room temperature, or design architectural plans that cut carbon emissions, or air travel that uses less fuel.
AI’s unique ability to make predictions based on the trends and patterns it spots in large quantities of data make it good at providing insights in areas that are otherwise uncertain, like climate change. AI’s predictive capabilities can be used in everything from climate modeling to the development of mitigation policies — allowing scientists to ask “what if” questions, and giving policymakers the knowledge they need to to weigh the costs and benefits of a particular strategy.
AI Has a Huge Climate Change Problem, E Glover 2023
Technology is ushering in profound changes; altering how we see the world and each other, how we function, how we relate. In 2011, philosopher Mark Poster wrote in the Introduction to Vilem Flusser’s extraordinarily perspicacious Into the Technical Universe of Images that studies into these fundamentally changed relations are urgent. It certainly feels urgent to me. In the same book, Vlusser writes that “To turn a technical question into a political one, it must be torn from the technician’s hands. Technology has become too serious a matter to be left to technicians”. It’s this kind of statement that keeps me going when I feel my resolve wavering for any number of reasons, from insecurity about my contributions to concerns about the carbon costs of technology-use. And as you have no doubt gleaned by now, I have not really conversed as I did in previous posts with the machine. And so this final interview is not an interview at all – because even I don’t trust it to provide honest answers about its carbon footprint which avoid slipping into propaganda.
Finally, I tried using Google’s Bard while writing this post, but when asked for “graphs showing the cost of AI in terms of energy consumption”, it replied “I’m just a language model, so I can’t help you with that”, potentially undermining everything I’ve just said. It’s tempting to imagine Bard being coyly dishonest. “What me? Gorging on energy like a large industrialised country? Never!”
Refs:
Agamben. G. (2009) What is an Apparatus? and Other Essays, T. D. Kishik and S Pedantella, Meridan
Braidotti, R. (2013) The posthuman. Cambridge, UK ; Malden, MA, USA: Polity Press.
Davis, B. (2022) Art in the After-Culture Capitalist Crisis and Cultural Strategy. La Vergne: Haymarket Books.
Flusser, V. (2011) Into the universe of technical images. Minneapolis: University of Minnesota Press.
Glover, E. (2023) AI and Climate Change | Built In At: https://builtin.com/artificial-intelligence/ai-climate-change-dilemma (Accessed 26/08/2023).
MacFarquhar, L. (2018) ‘The Mind-Expanding Ideas of Andy Clark’ In: The New Yorker 26/03/2018 At: https://www.newyorker.com/magazine/2018/04/02/the-mind-expanding-ideas-of-andy-clark (Accessed 30/05/2022).
Poster, M (2011) Introduction to V. Flusser, Into the universe of technical images. Minneapolis: University of Minnesota Press.
The National Gallery, London (2023) Art and AI podcasts | The AI Gallery | National Gallery, London. Available at: https://www.nationalgallery.org.uk/national-gallery-x/the-ai-gallery/art-and-ai-podcasts. (Accessed 12/08/23)
Schwaller, F (2023) Generative-AI: Dreaming up proteins – DW – 07/27/2023 (s.d.) At: https://www.dw.com/en/generative-ai-inventing-proteins-is-changing-medicine/a-66356415 (Accessed 26/08/2023).