AI has great potential for social good
Earlier this year, I attended EmTech Digital, a leading forum on AI and the latest stage of technology organised by the MIT Technology Review. Interestingly, the major emphasis this year was on ethical AI and the use of AI to solve social issues and to create a positive impact.
Among recent movements in the tech community, Google has launched a programme called AI for Social Good, to provide funding and technical advice for AI projects with social impact. IBM also puts in a similar effort in its Science for Social Good project.
For the first time in history, AI no longer belongs just to a small circle of experts. Many companies are democratising the technology by building tools to make the complex technology accessible to all. For example, Microsoft is developing a tool that allows users to drag and drop to build a sophisticated Machine Learning model. AWS, the cloud services arm of Amazon, has also set goals to build cloud infrastructure and add capabilities to simplify the AI building process. Wider adoption will surely take place.
Today is indeed the prime time to expand the use of AI to tackle social problems.
Applications of AI for social issues are wide-ranging. For example, projects on Google's AI for Social Good programme range from predicting a heart attack to protecting wildlife. A paper by McKinsey Global Institute curates a list of over 100 use cases of AI for social good with most use cases in the areas of health and hunger such as early-stage diagnosis and food and vaccine distribution optimisation.
In Thailand, I have witnessed recent explosive growth in terms of talent pool, resources, as well as the potential of AI applications. Companies and government organisations alike have been integrating AI use cases into their daily operations and using them for commercialisation purposes. More attention has been paid to social causes as well, as seen in the growing number in hackathons that target issues like education and agriculture.
However, a one-off solution emerged from hackathons that may not translate to actual large-scale problem-solving. There is still a missing link. To realise the full impact, we may need commitment and collaboration from multiple parties including private and public. The participation could be in a more passive form such as allowing data access. Imagine if we can tap into the massive amount of data from traffic cameras, real-time satellite images, and geolocation signals to help solve this long traffic programme facing Bangkokians every day.
For example, researchers at Carnegie Mellon University are working with Pittsburgh city officials to use image classification and object detection to optimise a traffic control system. They were able to reduce standby time for traffic by 40%, cutting travel times by as much as 25%, and also pollutant gases by 20%. Similar efforts are being made in a collaboration between the Alan Turing Institute in London and the Toyota Mobility Foundation in Japan. If this is possible in Bangkok, it could mean a lot less time spent idling at intersections. Perhaps, this could present a better way to solve the PM2.5 problem than giving away free masks.
Another major social issue where AI can contribute is inequality, whether in income, education or opportunities.
A few years back, the government launched a welfare card scheme, the so-called card for the poor, in an attempt to better target people for poverty-alleviation programmes. This yielded poor results with 14 million "poor" people turning up, in contrast to the hard statistics showing only 4-5 million.
Instead, we could use technology to do the job of targeting the needy. A team of researchers at Stanford University have used high-resolution satellite imagery and machine learning algorithms to identify poverty-stricken areas. This helps governments better target cheaper costs and ease of scaling.
Another use case of using AI to tackle inequality is the use of AI to help farmers, often at the lower rung of the social ladder. A start-up company, Ricult, has been using machine learning models on weather patterns and satellite imagery to improve crop yields for farmers.
Another set of researchers used similar sets of data to create weather insurance to help farmers cope with risk.
These are inspiring examples of how AI can be applied to tackle social issues. Aside from using the technology to do good, it is just as important to use it responsibly, especially for businesses.
We are no longer in a period when companies can do whatever it takes to "hack" growth. We have already witnessed the unintended consequences of new technology like the case of Facebook and other social media. Rather, we should focus more on sustainable innovation and responsible technology, the kind that instills trust, security and reliability.
With an ever-advancing frontier, technology has become extremely powerful. It is going to get more difficult to keep the wielders of this power in check with moral boundaries. Deepfake is one of those technologies that could be the tipping point towards an state of anarchy whereby no one can believe anything they hear or see. It is up to all of us in society to stay on the good side and keep making positive steps forward.
This is my final article for this column. Since 2012 when I started the column, it has been a great pleasure for me to express my opinions and receive comments from thoughtful, impassioned readers via emails and in person. I have learned tremendously from such interactions. Above all, I hope my articles have inspired lively debates over open markets, open technology and open polity in the open society we all deserve.
Sutapa Amornvivat, PhD, is CEO of SCB ABACUS, an AI-powered data analytics subsidiary of Siam Commercial Bank, where she previously headed the Economic Intelligence Centre and the Risk Analytics Division. She received a BA from Harvard and a PhD from MIT. Email: firstname.lastname@example.org
CEO of SCB ABACUS
Sutapa Amornvivat, PhD, is CEO of SCB ABACUS, an advanced data analytics company under Siam Commercial Bank, where she previously headed the Economic Intelligence Center and the Risk Analytics Division. She received a BA from Harvard and a PhD from MIT. Email: SCBabacus@scb.co.th