Artificial Intelligence: How can it unlock opportunities for people and the planet? (1/2)
TL;DR
Artificial Intelligence has vast potential to help meet climate change targets. However, today AI expertise is often too concentrated, both geographically and within select technology companies and academic institutions. Dr Priya Donti, the co-founder of an organisation working at the intersection of climate research and AI, has ideas for how the two can come together to create new opportunities across diverse teams, sectors, and countries.
Let’s start with some of the basics:
🧠 What is artificial intelligence?
Artificial Intelligence (AI) is the umbrella term for algorithms that perform complex tasks, often (though not always) those associated with human behaviour. Within this is a sub-field called ‘Machine Learning’ (ML), which is the ability of the computers to automatically interpret and/or learn patterns in large complex datasets.
⏱️ Why are we talking about it now?
It is not an exaggeration to say that AI is an inflection point in human development. Whilst we are just at the beginning of seeing what its potential will be, there is no doubt that the next decade will be transformative.
A BCG survey conducted in May 2022 found that 87% of global public- and private-sector leaders who are responsible for climate or AI topics believe that AI is a useful tool in the fight against climate change.
Before we dive in, It is important to recognise that AI is not uniquely helpful by default and is used in applications that add to climate change, such as advancing oil and gas exploration. Therefore the intention and design are critical to ensuring positive outcomes.
🌍 So how can it help with action on climate change today?
In a nutshell there are six core applications:
Data mining: gathering and analysing complex data to suggest actionable insights
Accelerated experimentation: using findings from historical experiments to suggest next steps
Speeding up time-intensive simulations: to help with future predictions
Forecasting: predicting future events in the short and long-term to help assemble the right teams
System optimising: finding opportunities for operational efficiency
Predictive maintenance: detecting faults early to save on damage down the line
Source: Aligning artificial intelligence with climate change mitigation
AI has many ways to advance manual work being done on climate change mitigation and adaptation. Here’s a pick of some really exciting ideas:
👀 Data mining and insights
Climate TRACE is a coalition of organisations that have built an approach to emissions monitoring that combines carbon emissions data from more than 300 satellites and 11,000 sensors, to help with international climate negotiations
Mapping the Andean Amazon Project (MAAP) uses remote sensing to track and prevent deforestation in real time
By leveraging public enthusiasm for ornithology (birdwatching), eBird has logged more than 140 million observations and used them for population and migration studies
🔋 Accelerated experimentation
US-based startup Aionics uses machine learning to analyse the outcomes of past experiments on battery designs, and then recommend which designs to try out next, enabling on average a 10x reduction in the number of experiments
📈 Speeding up time-intensive simulations
Although early in development, AI algorithms are being used to fill data gaps on how and when climate change will impact financial assets, affecting both individual companies and financial markets. It will also help to assess whether the physical and transition risks disclosed by companies are accurate.
Analyzing how large-scale ecosystems like forests might react to different potential climate futures. These predictions will aid manual work on the ground to try to protect vulnerable biomes
🔭 Forecasting
Google has several projects on hazard forecasting and disaster relief: in areas prone to floods, wildfires, hurricanes and pests, to help citizens prepare and help governments with disaster relief operations
Kuzi, a tool developed by the Kenyan company Selina Wamucii, provides small-holder farmers with early warning of locust outbreaks, which are being exacerbated by climate change
The ICENET project forecasts how ice is melting. Amongst other things, understanding sea-ice this helps indigenous communities adapt, and help scientists more accurately predicts sea-level rise. Glacier analysis is critical for predicting water flows and shortages, such as in the Himalayan region where 1.4 billion people depend on glacial basins for drinking water
IceNet’s forecasts for July, August, and September 2020 at a 1-month lead time
Source: Seasonal Arctic sea ice forecasting with probabilistic deep learning
🚢 System optimization
The UK National Grid Electricity System Operator and Open Climate Fix work together to forecast national electricity demand and aid with infrastructure planning. This enables more efficient power supply, dynamic pricing and trading, and support with integrating renewable energy sources
Infrastructure company Arup has created Neuron, a tool which collects real-time data on heating, ventilation, and air conditioning in buildings, and uses AI to optimise energy use
Multiple companies are working on optimising logistic routes and consolidating freight shipments to reduce the number of trips by ship, air, or road, and therefore reducing carbon emissions
🛠️ Predictive maintenance
In machinery this involves predicting where wear and tear will occur, or where there might be breakages in electrical, water, and transportation infrastructure
AI can also help optimize recycling processes and waste, sorting for energy-intensive materials and reducing the virgin materials needed to be created
❓ What next?
These are all exciting individual projects, but we need more wide-spread understanding of this technology. The AI landscape is still highly fractured, with expertise concentrated in the Global North in select (mostly private sector) groups.
Today 78% of leaders feel there is insufficient access to AI expertise across the board. Continuing on this trajectory risks a future where a small set of companies have built up a closed monopoly of data and expertise, whereas there is a widespread lack of AI capacity across the majority of organisations.
Source: AI for the Planet by BCG
So how can we better combine climate change and artificial intelligence, and accelerate climate action? Stay tuned for an interview with Dr Priya Donti, the co-founder of Climate Change AI and a leading voice in this space.
Further reading & watching:
Tackling Climate Change with Machine Learning (a hefty but ground-breaking analysis of the possibilities in this space)
Watch this presentation by The Alan Turing Institute on IceNet: Deep learning framework for sea ice forecasting