It is becoming increasingly clear that rapid advances in algorithmic systems associated with artificial intelligence (such as machine learning, intelligent infrastructure, the Internet of Things) are likely to pose difficult challenges to governance and policy in multiple sectors such as medicine, finance, policing, urban planning, transport, and energy systems. In practically all modern domains, such technologies increasingly serve to impose order, hierarchize needs, allocate resources, and impact the distribution of wealth and opportunity. Algorithmic systems presently interact so closely with human decisions and behavior that these systems now influence many aspects of modern life in almost invisible ways. Questions such as who is policed, where it is lit, how much funding a community receives, and how jobs are allocated, are all becoming subject to the hidden hand of algorithms. Yet, much like Goethe’s (and Disney’s) sorcerer's apprentice, we may not be able to control or even understand the tools that shape our contemporary life. We may not even know the location or extent of failures caused by AI until damage has cascaded into catastrophe.
One often ignored fact is that algorithmic systems of different degrees of sophistication are already important today in the environmental and ecological domain. This includes social media mediated filtering of environmental information online, environmental monitoring systems, climate change modeling, energy distribution, and urban and landscape planning, just to mention a few. Advanced algorithmic systems and robotics are also making rapid progress in the agricultural sector, for example through “precision agriculture” investments in Europe and in Asia, and in technologies for marine exploration and exploitation. It would be a mistake to assume that these applications will remain flawless in an environment of rapidly changing ecological circumstances.
This means that not only our monitoring of the present, but also our forecasting of the future is controlled by mechanisms whose collective behavior we do not understand. The possibility of a disastrous accident in agriculture is a critical issue in a world where we remain “three meals away from chaos.” Moreover, as we come to depend on constant monitoring of the environment and subsequent policy choices, the very means by which we make the world legible need to be better understood.
The Stockholm Resilience Center, Beijer Institute, and PIIRS GRS Research Community have teamed up to organize a seminar and workshop to adress these growing concerns in the world of artificial intelligence. The central focus will be on the way emerging AI-systems shape the way we perceive and respond to environmental change, and how it could fundamentally alter the ways in which humans modify ecosystems around the world and impact human wellbeing. With the goal of fostering fruitful discussion and eventually a published work, this multidisciplinary workshop will bring together scholars from computer sciences, ecology, political science (amongst others), and actors from the IT-sector where these systems are currently being developed and tested on the ground.
Preliminary list of Participants:
In advance of our workshop this winter in Princeton, we will collect responses from a pre-workshop online survey in order to share the academic backgrounds—and thoughts on the workshop topic—of all attendees in advance. We have aggregated all of the responses into a single document, which is available at the link below, and which we will also distribute to all attendees at the workshop.
If you have would like to change any responses in the above document, please email us at gsr@princeton.edu.
Link to more detailed workshop schedule and agenda
Day 1 (Friday) will begin at 8:30 AM with coffee and a light breakfast in Louis A. Simpson Intl. Building (20 Washington Road), Room A71
Day 2 (Saturday morning) will begin at 8:30 with coffee and a light breakfast in Wallace Hall, Room 300. The day will start with a final panel followed by a structured workshop to begin discussing what published works could come from this event.
Travel: The workshop organizers will provide travel and hotel accommodations in Princeton. Please see travel and hotel information below and email Thayer Patterson and Peter Callahan at gsr@princeton.edu if you have any questions.
Most participants will fly into Newark Airport (EWR), which is an hour by train from Princeton. Trenton, Philadelphia, JFK, and LaGuardia airports are other nearby options.
The train that travels from Newark Airport to Princeton’s campus is operated by New Jersey Transit. Once you arrive at Newark Airport, you can take an airport tram (the “Airtrain”) from your terminal to the Newark Airport train station, where you will purchase a roundtrip ticket to “Princeton.” From there you’ll take the New Jersey Transit Northeast Corridor line towards Trenton, and disembark at Princeton Junction. Upon arrival, you will find a shuttle bus that travels the five minutes from Princeton Junction to the Princeton campus “Dinky Station” (the famous “Dinky” train from Princeton Junction to the Princeton campus is out of service until late January and has been replaced by a shuttle bus).
Conference participants will be staying at the Nassau Inn, located on Palmer Square in downtown Princeton, just across the street from the University. This hotel is a 10-minute walk through the Princeton campus from the Dinky Station, and puts everyone a short walk from the workshop location. Should you get hungry upon arrival, there is a small pub and restaurant in the Nassau Inn, as well as other restaurants on Palmer Square. If you arrive after midnight when the shuttle is no longer running, you may take a taxi/Uber to the Nassau Inn.
Link to Memo AI Workshop Memo
Distributed via email on November 8, 2018. Updated November 14th.
For more information about the Workshop or the Global Systemic Risk Research Community, please email the group coordinator, Thayer Patterson, at gsr@princeton.edu.