Let us imagine an Ecological Intelligence Agency, suggests Superflux - a way for AI to help us be good stewards of nature
From the beginning of AG, we’ve love the design agency Superflux’s capacity for the concrete imagineering of possible futures, both desirable and cautionary one. Here’s their latest speculation, completed for the UK Govt’s environmental agency Defra, on how to create an “ecological intelligence”, built from AIs and samples of nature.
Their brief was: “How can we transform what the freshwater system looks like, 20 years from now and beyond?” News-watchers will perhaps note that freshwater supplies in the UK might be immediately improved by water companies ceasing to cheaply dump sewage in rivers. But there’s certainly no harm in imagining an optimal system - where AIs can clearly help to detect misbehaviours.
Below are some excerpts from Superflux’s own site:
Vision & Concept
AI Foundational Models have considerable potential: the potential for radical positive change as well as the threat of existential risk comparable to nuclear war and pandemics (many silicon valley engineers have calculated a P(doom) of 10%).
AI continues to make today’s headlines but it’s been with us for a long long time; policing, recruiting, suggesting. Whilst we can speculate about where foundational models might go, we know of the extractive and exploitative operational practices that go into them, and the few who currently yield the power. Before we think about what AI can do, we need to rethink how we do AI.
Are there entirely alternate intelligences that can be nurtured here? How can we unite labour, climate and data justice in a way that is not extractive but mutually supportive and led by the commons? Could AI be an advocate for ecological health? These are the questions we began with.
To create a foundation for Ecological AI, we considered an alternate intelligence that does not claim unlimited access to knowledge systems, does not assume only one way to understand the world and does not perpetuate dominant extractive paradigms.
What of an intelligence that is accountable and accounts for a multitude of interconnected cosmologies and lived-experiences alongside data sets?
Pushing this further, we wanted to render scientific evidence meaningful through the poetic, as well as make the journey transparent and traceable, in turn amplifying the voice of actants often left out of analyses and decision making.
The result: The Ecological Intelligence Agency.
In a series of texts and videos, Superflux asks you to imagine what an EIA would do:
The Ecological Intelligence Agency: Aligning with Ecological Intelligence for a Flourishing Planet
The Ecological Intelligence Agency is an autonomous inter-departmental government agency that encompasses an assemblage of localised AI models all of which advocate for ecological flourishing.
Co-developed with and informed by the wisdom of local stewards, the EIA has been intentionally developed for inter-relational use, bridging sectors, departments and communities to help build a (partial) picture of and communicate our entangled interdependence with the world around us.
Taking you on a journey across temporalities, data and poetics, the EIA encourages ecological perspectives for policy formation, regulation and implementation. Its intent is to make river health sense-able and help situate policies within wider contextual ecosystems.
One large concern with LLMs is that even those who built the systems do not know how they reach the answers that they do. The untraceability of LLM’s outputs raises problems for regulation, intellectual property and accountability.
An AI that is necessarily ecological will not conjure a traceless answer in an authoritative singular voice but advocates for a multitude of perspectives, giving voice to (and recognition of) unheard communities and our under consulted more-than-human companions.
We considered how an Ecological AI would draw connections and pursue reaction chains, whilst understanding the potential of gaps and its own limitations. Where the EIA encounters gaps in the knowledge it points to those who must be further consulted and involved in the decision making process.
In the words of mathematician and philosopher Alfred Whitehead:
“A proposition is neither true nor false in itself, but it is rather a lure for feeling the world differently, and its efficacy is determined by what it does, by what it makes possible to feel and think”.
Surfacing a plurality of voices in a mytho-poetic chorus, The EIA invites you to listen, feel and think with the situated wisdom of the waters.
This produces three video simulations, in which Superflux use LLM AI’s to speak in the voice of the River Roding, generating poetry from that voice - “lamenting the quiet cascade of forever chemicals, to imagining a harmonious future”.
The first video (embedded at the top of the post) looks at “pollution in River Roding, and focused on the impact of forever chemicals on ecosystems through cascades such as food chains”.
The second video below “centred on sewage issues, looking at the wider implications of large infrastructure projects and the effects on the river such as algal blooms”:
The third scenario “considered increased flooding risks, and speculated on ways of living alongside natural water flow patterns”: