The Association for the Advancement of Artificial Intelligence (AAAI) held a unique gathering at Stanford University in March 2017: “AI for Social Good.” Organizers were seeking to explore real-world applications of artificial intelligence for issues like urban planning, health care, social justice, sustainability, and national security. They sought to “bring together AI researchers and researchers/practitioners/experts/policy makers from a wide variety of domains” to have a dialogue. Speaking as an environmental psychologist, I was pleased to have a brief paper accepted for the meeting: “Can Artificial Intelligence have Ecological Intelligence?” Unfortunately, I was unable to attend and present it in person. But, I wanted to share some of the main ideas, and you can get into the details in my proposal below.
In case you haven’t been keeping track, artificial intelligence and its rapid development is considered one of the most important issues in the world right now. So, I think we are lucky that there are organizations like AAAI having these discussions. The people who are developing AI need to hear from us.
My presentation proposed that we take what we know about how humans develop their “environmental identities” (their sense of personal identity in relation to nature and the natural environment), and the moral responsibility that flows from this, and apply these insights when we create intelligent computer systems. In science fiction stereotypes, computers or artificial intelligence are typically portrayed as a threat to nature and environmental sustainability. But, the fact is, that need not be so. It’s possible that we can set up machine learning so that artificially intelligent systems are as good or even better than us in terms of nature ethics and the ability to balance the complex needs of people and the planet. I noted practical transportation examples that might be:
“… as mundane as having cars that not only track their miles per gallon but also their current and cumulative carbon footprints, or their contributions to local air quality (i.e., such as by monitoring their idling). These systems may be as sophisticated as navigation equipment that helps airliners or ocean ships to avoid collisions with migrating species.”
But here’s the real challenge with thinking about AI: How can we think as big as possible about how beneficial AI can be. Not only can intelligent systems track more data in real time, they can also reach back into the past and make powerful estimates about the future. AI can be “self-aware” in the way that we mere humans can only strive to be. As I noted,
“What would it mean for machines and artificial intelligences to become aware of their own role in the natural environment as ethical and decision-making agents, their own provenance and product history, and their ability to promote sustainability or species biodiversity and conservation—or their “environmental identity” and “ecological intelligence?” Can this be leveraged to augment and improve human decision making and overcome common cognitive biases, such as those that lead to “tragedy of the commons” issues?”
I am not a technologist by training. But my research tells me capabilities like having machines know the story of their supply chain is possible right now, even where the raw materials were mined that that make up their components, and how these can best be recycled. The only thing holding us back is our creativity and ability to ask for what we need. I am also not a simplistic “tech booster” either. I know there are trade-offs with all new technologies and that every innovation has a shadow side. But remember AI is made by people like you and me. In terms of humans, AI and nature, I am suggesting we focus on the positive and dream big.
— Dr. Thomas Doherty
My AI for Social Good (AISOC17) paper proposal:
“Can Artificial Intelligence have Ecological Intelligence?“
This theoretical paper and discussion explores the role of environmental ethics and decision-making in the development of artificial intelligence (AI) and the benefits of computer-enhanced human decision-making and behavioral systems. To spur thinking and discussion, the author will apply commonly used theoretical and empirical concepts from environmental psychology—that describe humans’ beliefs and values regarding nature, the natural world, other species, sustainability, and issues like climate change—into an AI context.
Environmental psychologists and other social scientists have identified a number of concepts that describe, measure, and give language to people’s values and connections to nature and the natural world. These include “environmental identity” (one’s sense of personal identity in relation to nature), experiences of “ecological self” (immersion, flow, or interbeing with the natural world), and “ecological intelligence” (understanding of one’s behaviors and purchases in relation to the environment and sustainability, such as understanding the supply chains of commonly used products). Privileges or disparities related to access to healthy natural settings and experiences, or injustices regarding exposure to environmental issues is also well studied.
What would it mean for machines and artificial intelligences to become aware of their own role in the natural environment as ethical and decision-making agents, their own provenance and product history, and their ability to promote sustainability or species biodiversity and conservation– or their “environmental identity” and “ecological intelligence?” Can this be leveraged to augment and improve human decision making and overcome common cognitive biases, such as those that lead to “tragedy of the commons” issues?
The author will provide some basic empirical support for the constructs of environmental identity and ecological self, and commonly used models of environmental decision-making, such as Values-Beliefs-Norms theory. These can inspire intelligent systems design, pervasive computing, and decision algorithms that can promote a variety of environmentally-significant goals. These systems might be as mundane as having cars that not only track their miles per gallon but also their current and cumulative carbon footprints, or their contributions to local air quality (i.e., such as by monitoring their idling). These systems may be as sophisticated as navigation equipment that helps airliners or ocean ships to avoid collisions with migrating species. In terms of direct human benefits, research on restorative “living buildings” and benefits of connection with green views and natural spaces can be harnessed in terms of better user interface for “smart” sustainable buildings and neighborhoods. These methods might be made available for the population health of the general public, or used directly in healthcare settings as therapies (e.g., tracking time of exposure to healing gardens, or indoor air quality and light, and patient outcomes).
AI has the potential to link and monitor a number of nested and interconnected human and natural systems —in real time, and over time (including past data and future forecasting). AI might operationalize a number of long-standing environmental goals and dreams that have traditionally been seen as conceptual, metaphoric or philosophical.
The potential societal benefits of ecological AI are immense. These help to counter the narrative that AI and machine learning ultimately leads to environmental destruction and dystopia. A more reality-based discussion will recognize that humans tend to project both their positive and negative ecological behaviors onto technologies. Rapid technological advances such as AI can help realize goals for pro-social and ecological benefits and also expand human conceptions about what is possible in terms of sustainability.
Bratton, B. (2016). The Stack: On Software and Sovereignty. MIT Press.
Clayton, S. (2003). Environmental Identity: A Conceptual and Operational Definition (pp.45-66) in S. Clayton & S. Opotow Eds. Identity and the Natural Environment. Cambridge, MA: MIT Press.
Doherty, T. J. & Chen, A. (2016). Improving Human Functioning: Ecotherapy and Environmental Health Approaches. In R. Gifford (Ed.). Research Methods in Environmental Psychology. John Wiley & Sons.
Doherty, T. J. (2015). Mental Health Impacts. In J. Patz & B. S. Levy (Eds.) Climate Change and Public Health. Oxford University Press
Doherty, T. (2015). “Psychology & Nature” (Video File). Retrieved from: http://bit.ly/psychnature.
Fisher, D. (2011). Computing and AI for a Sustainable Future. IEEE Intelligent Systems, 26.
Foth, M., Paulos, E., Satchell, C. & Dourish, P. (2009) Pervasive computing and environmental sustainability: Two conference workshops. Pervasive Computing, 8, 78-81.
Gifford, R. (2014) Environmental Psychology Matters. Annual Review of Psychology, 65, 541- 579
Goleman, D. (2010). Ecological Intelligence. New York: Crown Business.
Ramchurn, S. D., Perukrishnen Vytelingum, P., Alex Rogers, A. & Jenning, N. R.
(2012). Putting the ‘Smarts’ into the Smart Grid: A Grand Challenge for Artificial Intelligence. Communications of the ACM, 55, 4., 86-97.
Sandler, R. & Pezzullo, P. C. (Eds.). Environmental Justice and Environmentalism: The Social Justice Challenge to the Environmental Movement. Cambridge, MA: The MIT Press.
Saunders, C., Brook, A. & Myers, M. (2006). Using Psychology to Save Biodiversity and Human Well-Being. Conservation Biology, 20, 702–705.