Artificial Intelligence is frequently in the headlines these days, sometimes portrayed in apocalyptic terms as the technology that will take over our jobs or even our lives. But what if it could also become a valuable tool in the worldwide efforts to achieve the UN’s Sustainable Development Goals?

Already, AI capabilities are being used in various ways to further societal goals. After the passage of Hurricane Harvey in 2017, many streets in downtown Houston were flooded and impassible—but others weren’t. An AI application that combined satellite imagery with object detection software enabled rescue workers to identify safe escape routes for those trapped by the rising waters.

AI-powered object detection is also at the core of a new application that could bring relief to the estimated 250 million people worldwide who are visually impaired, most of whom live in developing countries. The free Microsoft app, which works through a smartphone, uses AI to recognize friends and describe people and specific objects such as currency bills.

These are just two cases of AI being used for social good—and it’s just the start. We at McKinsey Global Institute have collected about 160 cases of AI’s actual or potential uses for noncommercial benefit of society, and can see considerable potential for computer vision, natural language processing, even deep learning applied to traditional databases, as well as other AI capabilities, to make a very sizeable difference.

Our evolving library of cases already touches on all 17 of the SDGs. The goal of good health and well-being is particularly well represented in our use case library. For example, more than 400 million people worldwide afflicted with diabetes could in theory be helped by an AI-enabled wearable device that can already detect potential early signs of diabetes through heart rate sensor data. AI could also analyze mobile phone images to distinguish between a benign skin lesion and melanoma—potentially helping millions of rural dwellers who have no easy access to dermatologists.

While AI’s potential to make a real difference exists, realizing that potential will require a very considerable scaling up of its uses for social good. Only about one-third of the cases in our library are being applied in the field today and, even then, many remain small-scale tests.

Getting to the next level will require some decisive action by governments, NGOs, philanthropists, tech companies, and organizations that collects or generate significant amounts of data.

Two critical bottlenecks that will need to be overcome: problems with data accessibility and a chronic shortage of talent able to improve AI capabilities, improve models, and implement solutions.

And AI application does not come risk-free. Its tools and techniques can be misused by authorities and others—or be subject to unintentional misuse, sometimes harming the very people they are intended to help. For example, while several AI applications could be used for the 'Peace, justice, and strong institutions' (SDG 16), challenges around how to explain the outcomes of some complex AI models means the use of such models could be difficult to accept in such sensitive areas.

These obstacles are not unsurmountable. To improve data accessibility, for example, collectors and generators of data, whether governments or companies, will need to grant greater access to NGOs and others seeking to use the data for public service. The talent shortage may require some conscious acts of altruism by technology companies and others who employ highly skilled AI talent.

AI is not a silver bullet for all of humanity’s problems. But it has the potential to be a formidable tool in the toolkit. For that to happen, we need to encourage the development of applications, and push for them to be used at scale in a responsible and thoughtful manner.

Michael Chui is a partner at the McKinsey Global Institute, the business and economics research arm of McKinsey and Company, where Rita Chung is an MGI fellow and engagement manager. Ashley van Heteren is an expert associate principal and a leader of McKinsey’s Noble Intelligence venture, which uses artificial intelligence to solve humanitarian challenges, including helping the unemployed, combating disease outbreaks, and fighting human trafficking, among other causes. They are co-authors of Notes from the AI frontier: Applying AI for social good.