How are Siri, Alexa, Watson, and the significant others, shaping the sustainability discourse?

The study of artificial intelligence began way back in 1956 when a group of researchers got together in New Hampshire (Dartmouth Summer Research Project on Artificial Intelligence hosted by John McCarthy and Marvin Minsky) to discuss how machines could perform “intelligent actions”. Since then the landscape has gradually evolved, thanks to technological changes such as increased computational power and new discoveries in neuroscience. Artificial intelligence (AI) is the simulation of human intelligence processes by machines, especially computer systems. These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions) and self-correction. Plainly put, AI can be divided into 3 categories. The first broad classification is artificial intelligence itself, defined broadly and covering all possible approaches to simulating intelligence. A subset of artificial intelligence is machine learning, which uses data and experience automatically to tune algorithms. Finally, a subset of machine learning is deep learning. Deep learning uses brain-inspired algorithms – neural networks – to simulate the learning process. Because of data explosion associated with social networks, connected sensors, and seismic oil exploration among others, machine learning has become quite popular. Among machine learning algorithms, deep learning can absorb the most data and has broken previous AI records, becoming the most promising approach to AI.

The concept of sustainability is one that has been around for as long as humans have: a concern for the future of our resources. Coined in German, the original term was Nachhaltigkeit, meaning ‘sustained yield’. It first appeared in a handbook of forestry published in 1713 and was used to mean never harvesting more than the forest can regenerate. The translated term appeared in English beginning in the mid-19th century. The term sustainability is derived from the Latin sustinere (tenere, to hold; sub, under). The most widely quoted definition of sustainability as a part of the concept sustainable development, that of the Brundtland Commission of the United Nations on March 20, 1987: "sustainable development is a development that meets the needs of the present without compromising the ability of future generations to meet their own needs". Sustainable development has three goals, set forth by the 2005 World Summit, and they are economic growth, social development, and environmental protection.

On 1 January 2016, the 17 Sustainable Development Goals (SDGs) of the 2030 Agenda for Sustainable Development (adopted by world leaders in September 2015 at a historic UN Summit) officially came into force. These goals are unique in that they call for action by all countries, poor, rich and middle-income to promote prosperity while protecting the planet. They recognize that ending poverty must go together with strategies that build economic growth and addresses a range of social needs including education, health, social protection, and job opportunities while tackling climate change and environmental protection. The UN SDGs provide a lens for the challenges facing humanity.

As the world’s current population of around 7 billion grows to 9.8 billion by 2050, it will increase the demand for food, materials, transport, and energy, further increasing the risk of environmental degradation and affecting human health, livelihoods, and security. Keeping this in mind, I would like to list a few AI applications (or use cases) in each of the UN SDGs. For easy classification, I have grouped the 17 SDGs under the 3 broad pillars of sustainability: economic, environmental and social.

ECONOMIC

Goal 8: Promote inclusive and sustainable economic growth, employment and decent work for all

The founders of Belong.co said they shared a common challenge – that of hiring talent to build teams. As they spoke to several talented but discontented people, they said it became clear that there was a mismatch between passionate people and the places where they worked. This stood in the way of them finding satisfaction in what they did and unlocking their true potential. That’s the problem Belong.co says it is solving. Belong.co’s talent search suite, Belong Hire. claims to have saved 17 hours per recruiter each week, translating to $100,000 saved weekly. It consists of various tools, including Curation Engine: Curates talent data sourced from 90 different professional networks, communities and forums to offer hiring managers an untapped pool of ideal candidates. Adaptive Search: Uses machine learning and predictive analytics to uncover people like the most successful candidates based on the company’s sourcing and interview patterns. Belong Pick: While Adaptive Search uses historical sourcing and hiring data, Belong Picks analyzes current and past employee data to identify talent who are most like the high-performing employees. Job Description Scanner: The company claims this automatically understands a job description and generates relevant, skilled candidates. Belong.co clients include Adobe, Accenture, PayPal, and Cisco.

Goal 9: Build resilient infrastructure, promote sustainable industrialization and foster innovation

Gartner estimates that over 170,000 commercial drones will be sold this year, a 58 percent increase since 2016. Goldman Sachs predicts that businesses and the civil government will spend $13 billion on drones between 2016-2020. PwC projects that globally, drone solutions have the high potential to replace $127 billion in current business services and labor costs across multiple industries. They assume drones could have the biggest impact on infrastructure, agriculture, and transportation. The use of drones to inspect existing infrastructure can be cheaper, faster, and importantly safer. For example, the New York Power Authority tested using drones to inspect an ice boom near Lake Erie. Inspecting one of these ice booms normally costs $3,500 to send a helicopter or $3,300 to send a boat to perform the task, but it cost less than $300 to deploy a drone. Several energy companies like Southern Company and Duke Energy are using drones to inspect power lines, power plants, and storm damage. SunPower is using drones to improve solar farms. Similarly, start-ups such as SkySpecs are using drones to perform inspections of wind turbines which used to take hours in just minutes. or catch mistakes.

Goal 10: Reduce inequality within and among countries

Facebook initiated a project to use drones not just as a temporary replacement for current infrastructure but as a permanent alternative. Facebook called this project, Aquila, after its massive solar-powered drone that would stay in the air for months beaming the internet to remote or impoverished areas. The technology is still in the development phase, but the goal is to bring the internet to the “more of the 4 billion people who are not online” and therefore not using Facebook. Project Loon, a similar idea using high-altitude balloons rather than drones. It is being worked on by X, the experimental arm of Google parent company Alphabet. Both these projects are presently halted but it sure gave us a glimpse into a possibility that could be well explored in the future.

Goal 12: Ensure sustainable consumption and production patterns

Data from drones, remote sensors, satellites, and smart farm equipment provides farmers with valuable real-time information on soil, crop health, and weather conditions. This intelligence helps farmers make smarter decisions on where to grow crops, how to optimize crop rotations, and when to sow, compost, and harvest those crops. For example, some agri-tech solutions analyze images to determine when the fruit is ready to be picked. Others include algorithms that identify microbes that promote crop growth without synthetic fertilizers. Data-driven software and AI solutions can help farmers manage their work more effectively by providing outcomes for regenerative agricultural practices without expensive and time-consuming field trials. AI can also help farmers at the outset by designing out avoidable food waste and preventing edible food from being thrown away. Farm-based food supply chains can become more efficient using visual imagery technology during food inspections. AI-enabled tracking can help retailers sell food before it goes bad, and AI algorithms can forecast and predict sales to allow restaurants and retailers to more effectively connecting the supply to demand when ordering food, thereby reducing avoidable food waste. According to McKinsey, by using these techniques to design out food waste, AI can generate an estimated economic opportunity of up to $127 billion a year in 2030, calculated as growth in top-line revenue.

Goal 17: Revitalize the global partnership for sustainable development

Finance and Travel Chatbots: In 2018 Visa Canada announced their collaboration with Finn AI, a Vancouver-based fintech company. The collaboration allows Finn AI to use Visa’s APIs from their developer platform and to enhance the natural language processing technology within Visa’s chatbot software. Visa claims their customers will benefit from the improved functionality of their customer service chatbot. The chatbot can purportedly provide exchange rate information to customers landing in a new country as well as help the customer find the nearest ATM. Customers can purportedly chat with Visa’s Finn AI-powered virtual assistant to prepare their bank, mobile banking app, and debit card for travel. This could include making sure mobile app travel notifications are turned on before leaving. A customer may also disable a misplaced card to prevent fraudulent transactions by querying the chatbot to do so. According to a case study from Finn AI, the company helped Bank of Montreal launch a chatbot for their mobile app that improved their customer experience. Bank of Montreal wanted to improve its customer service by providing their customers with the ability to make bank appointments and ask financial questions when away from their bank. The case study states that by the time the solution was launched, Finn AI and Bank of Montreal had covered over 250 individual customer questions in there training. They also covered over 200,000 possible utterances which include sentence fragments and colloquial language. By the time the case study was published, the Bank of Montreal had already seen an improvement of their Facebook customer feedback scores by +1 point.

ENVIRONMENTAL

Goal 13: Take urgent action to combat climate change and its impacts

A new field of “Climate Informatics” is already blossoming, harnessing AI to fundamentally transform weather forecasting (including prediction of extreme events) and to improve understanding of the effects of climate change. New hybrid systems of rules and tools can use data and AI techniques to build a “Community Distributed Data Escrow” system that could enhance climate-based disaster preparation and response through coordination of emergency information capabilities. When a disaster strikes, predefined uses of data would be activated to equip first responders with better tools for understanding the local context and take precise action. For example, machine learning combined with natural language processing algorithms could identify the best station points and routes for distribution and evacuation, the amount of relief required and optimal relief-effort timetables. Here AI would work in combination with other Fourth Industrial Revolution technologies including drones and the Internet of Things. Deep reinforcement learning may one day be integrated into disaster simulations to determine optimal response strategies, like the way AI is currently being used to identify the best move in games like AlphaGo.

Goal 14: Conserve and sustainably use the oceans, seas and marine resources

Real-time monitoring with AI can improve decision-making in fields ranging from species management and protection to natural resource management to climate resilience. One early example is the Ocean Data Alliance, which has started to work together to develop and implement open-source solutions to provide the data needed for comprehensive monitoring of ocean resources, from satellites to data from ocean exploration technologies. Developed fully, this approach could allow decision-makers to use machine learning to monitor, predict and respond to changing conditions such as illegal fishing, a disease outbreak or a coral-bleaching event.

Goal 15: Sustainably manage forests, combat desertification, halt and reverse land degradation, halt biodiversity loss

The Amazonia Third Way initiative is developing the Earth Bank of Codes, a project to create an open, global public-good digital platform that registers nature’s assets, recording their spatial and temporal provenance and codifying the associated rights and obligations. (This helps to implement the Nagoya Protocol of the Convention on Biological Diversity.) A fusion of AI and complex systems analytics will be vital to bundling the biological, biomimetic and traditional-knowledge assets from a biodiversity hotspot to maximize economic and conservation value simultaneously. In addition, an AI-driven “biological search engine” will allow users to understand more fully the planet’s web of life, which could optimize scientific discovery, catalyze a myriad of bio-inspired innovations and improve conservation outcomes by creating new sources of economic value. AI techniques will include natural language processing, deep learning, computer vision, probabilistic programming and an array of statistical machine-learning techniques.

SOCIAL

Goal 1: End poverty in all its forms everywhere

FarmView, a multidisciplinary research team at Carnegie Mellon University that is developing automated, data-driven decision tools to increase the yield of sorghum, a drought- and heat-tolerant grain that thrives in famine-prone parts of the world. Sorghum is common in the Sub-Saharan African and Asian regions and can withstand extreme heat and drought. The program uses a robot, sensors, and a high-quality camera to take photos of sorghum’s grain head. On the back-end, AI technology looks at the photos and extracts information, such as the size of the grain head and the number and size of the seeds, then estimates the quality and ripeness of the crop. The process allows crop breeders to compare over 1,000 varieties of sorghum and make better decisions about planting, cultivating, and harvesting. The goal, as the university’s site states, is to help farmers develop plants that produce more food on fewer acres with less water. At Stanford University, researchers from the Sustainability and Artificial Intelligence Lab are using machine learning and remote-sensing data to predict crop yields, specifically, as it relates to soybeans. Stanford researchers are also working on locating areas of poverty, which can be difficult, as accurate and reliable data from impoverished regions are often scarce. In 2016, Stanford University economist Marshall Burke turned to daytime satellite images by using AI to fill informational gaps. To do this, Burke and his colleagues fed an algorithm both nighttime and daytime images from Uganda, Tanzania, Nigeria, Malawi, and Rwanda, all of which have household survey data available — and instructed the algorithm to find features in the daytime imagery that are predictive of places that are lit up at night. The model found numerous features that relate to such things as agricultural regions, bodies of water, and urban areas, but also various elements that were hard to interpret. In all, the algorithm could predict poverty 81 percent to 99 percent more accurately than a nightlight-only model. Satellite imagery and AI could also be used to identify areas that require the most aid.

Goal 2: End hunger, achieve food security and improved nutrition and promote sustainable agriculture

Drones are a natural fit with the world of agriculture, where farmers can benefit from real-time information about large tracts of land. Drones can help track almost everything including water use, crop health, heat signatures, and soil analysis. Expensive aerial surveillance that could previously only be done occasionally with planes can now be completed weekly or even daily with drones that cost only hundreds of dollars. Back in 2015, the American Farm Bureau Federation released a study that found for farmers using drones as a service, the average ROI is “$12 per acre for corn, $2.60 per acre for soybeans, and $2.30 per acre for wheat.” Thanks to the strong ROI, many predict the use of drones for farming will continue to grow. WinterGreen Research estimates the current agricultural drone market at $494 million and expects it will increase to $3.69 billion by 2022. Several new start-ups and existing agriculture players are investing heavily in agriculture drones and the software needed to analyze the raw data. For example, John Deere has partnered with Sentera to provide its consumers access to their AgVault? Software and agriculture scouting drones. In just one season, Sentra equipment performed 8,000 flights gathering 175 terabytes of data. Drones are allowing farmers to gather so much data about their fields that managing all this data has created an important business opportunity. Similarly, Dupont’s agribusiness division, Dupont Pioneer, has invested in PrecisionHawk to help “farmers to increase profitability and sustainability through data-driven insights.” PrecisionHawk offers a fully integrated product for all tiers of businesses. It can provide the drones, the sensors, the software, the analysis, the pilot, the insurance, and the regulatory filings.

Goal 3: Ensure healthy lives and promote well-being for all at all ages

The Machine Learning in Healthcare: Expert Consensus survey published in January 2019 by Emerj, noted that over 50% of respondents believed that AI will be ubiquitous in healthcare by 2025. On an open-ended question, over 25% of respondents stated that they believe that AI will be nearly ubiquitous in the healthcare setting by 2025. The same survey also observed that “Decision support systems” was ranked as the most likely application to be improved by AI (for improving patient outcomes), with an average 4.15 score on a 1-5 scale and “Improving Health Outcomes” was rated as the factor of highest importance in getting a customer to buy an AI product, rated a 4.2 on a 1-5 scale. “Saving Money” was rated a close second with a 4.1 score. Telehealth (or Telemedicine) is a growing sector of the healthcare industry which has steadily gained traction and formed a profitable sector, according to Transparency Market Research. The market research firm projects that total US revenue will hit $19.5 billion in 2025 up from $6 billion in 2016. the majority of AI use cases and emerging applications for telehealth appear to fall into two main categories: firstly, Virtual Consultations: When companies develop platforms for video consultations between patients and medical professionals, while providers use machine learning to help analyze clinical data in a patient’s electronic health or medical record (EHR/EMR) to provide patient care recommendations, e.g. Lemonaid Health, InfiniteMD and secondly, Diagnostic Support: When companies develop machine learning algorithms for chatbots to recommend a diagnosis based on symptoms and patient health data, e.g. HealthTap, Ada Health. 

Goal 4: Ensure inclusive and quality education for all and promote lifelong learning

Education is a domain largely ruled by human-to-human interaction, and integration of AI has been slower to develop the necessary human-like attributes of responsiveness, adaptability, and understanding. Yet there are plenty of areas where AI’s inherent strengths help fill high-need “gaps” in learning and teaching. Carnegie Learning’s “Mika” software, uses cognitive science and AI technologies to provide personalized tutoring and real-time feedback for post-secondary education students, particularly incoming college freshman who would otherwise need remedial courses. Carnegie states the cost of such remedial learning as costing colleges $6.7 billion annually, with only a 33% success rate for math courses. The University of Southern California (USC) Institute for Creative Technologies is a pioneer in creating smart virtual environments and applications that draw on AI, 3-D gaming, and computer animation to develop authentic virtual characters and realistic social interactions. USC researchers have several ongoing projects in the space that hint at applications to come over the next two decades.

Goal 5: Achieve gender equality and empower all women and girls

According to the World Economic Forum’s 2018 Global Gender Gap Report, only 22% of AI professionals globally are female compared to 78% who are male. 95% of all customer interactions will be driven by AI by 2025, with consumers unable to differentiate bots from human workers via online chats and over the phone. Undoubtedly, the rise of AI will improve our lives in many respects, but it also presents serious risks. AI relies on algorithms that learn from real-world data, so it can inadvertently reinforce existing social biases. Gartner predicts that by 2022, 85% of AI projects will deliver erroneous outcomes due to bias in data, algorithms or the teams responsible for managing them. A study shows how images that are used to train image-recognition software amplify gender biases. For instance, the careers platform LinkedIn, had an issue where highly-paid jobs were not displayed as frequently for searches by women as they were for men because of the way its algorithms were written. The initial users of the site’s job search function were predominantly male for these high-paying jobs, so it just ended up proposing these jobs to men – thereby simply reinforcing the bias against women. One study found a similar issue with Google. In another research study, 2 large image collections – including one supported by Microsoft and Facebook – were found to display predictable gender biases in photos of everyday scenes such as sport and cooking. Images of shopping and washing were linked to women, while coaching and shooting were tied to men. If a photo set generally associates women with housework, software trained by studying those photos and their labels create an even stronger association with it.

Goal 6: Ensure access to water and sanitation for all

Machine and deep learning could enable a step-change in the optimization of water-resource management. Increasingly, AI has the potential to create distributed “off-grid” water resources, analogous to decentralized energy systems. Household smart meters can produce large volumes of data that can be used to predict water flows, spot inconsistencies and check leaks. The next stage will be to combine machine learning, the Internet of Things and blockchain to create a truly decentralized water system, where local resources and closed-loop water recycling gain value. Water resources could even be traded via blockchain. Furthermore, machine learning, predictive modeling, and robotics can be combined to transform current approaches to building and managing water infrastructure and to accelerate innovation in environmental engineering. Rivers, for example, could be engineered to autonomously adjust their own sediment flows. Coupled with AI-informed pricing, such approaches could optimize water usage and drive behavior change by providing incentives for water conservation.

Goal 7: Ensure access to affordable, reliable, sustainable and modern energy for all

In the energy grid, the application of machine learning, including deep learning, is increasingly widespread in industry. For the environment, the use of AI to make distributed energy possible at scale is critical for decarbonizing the power grid, expanding the use of (and market for) renewables and increasing energy efficiency. AI can enhance the predictability of demand and supply for renewables, improve energy storage and load management, assist in the integration and reliability of renewables and enable dynamic pricing and trading, creating market incentives. AI-capable “virtual power plants” (VPPs) can integrate, aggregate, and optimize the use of solar panels, microgrids, energy storage installations, and other facilities. Distributed energy grids may also be extended to incorporate new sources such as solar spray or paint-coated infrastructure of vehicles, and to allow AI-enabled “solar roads” to expand, connect and optimize the grid further. In solar roads, for example, AI could allow a road to learn to heat up to melt snow, or to adjust traffic lanes based on vehicle flow.

Goal 11: Make cities inclusive, safe, resilient and sustainable

Cities aren’t just where the effects of climate change may be felt, but also where we can find solutions. As engines of today’s global economic growth, cities are responsible for more than 80 percent of the world’s GDP. But cities are also the primary drivers of pollution, consuming more than two-thirds of global energy and emitting more than 70 percent of global greenhouse gases. AI, smart meters and the Internet of Things can also help forecast and optimize urban energy generation and demand – both city-wide and at the level of individual homes and buildings. Real-time AI-optimized energy efficiency can have an immediate and substantial impact on energy consumption, Google, for example, cut power use in its data-centers by 40% by using DeepMind’s reinforcement learning algorithms to optimize cooling. AI-enabled smart grids will also be critical for fast-growing emerging cities, and are in fact already being piloted, from Brazil to the Philippines.

Goal 16: Promote just, peaceful and inclusive societies

In early 2017 AT&T successfully tested their comically named Flying COW (Cell on Wings). The drone functions as a small replacement cell tower that can be deployed at disaster scenes or during big events to spread the load placed on static towers. Real-time aerial inspection allows for the ability to quickly survey sites. Major companies like Amazon and Facebook are investing significantly in the use of drones for delivery and expanding internet access, but for the most part, these applications are in the testing phase. There are companies like Zipline, which is currently using aerial drones to make medical deliveries in Rwanda, but that represents a minor application compared to what some companies envision.