A Google report assesses how AI can best serve humanity13. September 2019
A Google report assesses how AI can best serve humanity
New York, 13.9.2019
Google.org today publishes a report describing how start-ups, non-profit organizations, government leaders and scientists are using machine learning to solve some of mankind’s most important problems. The evaluation had analyses of over 2600 participants from 119 countries. Google AI Impact Challenge had provided 25 million dollars for the tender.
To find out how human-centric companies use AI, Google used natural language processing techniques and cluster analysis. The AI analysis was complemented by interviews with applicants and recipients of Impact Challenge. The report now published is apparently the first ever produced by a technology giant.
Participants included the Colegio Mayor de Nuestra Señora del Rosario in Colombia. Using computer vision and satellite imagery, it investigated illegal mining operations that are known to contaminate local drinking water.
New York University and the Fire Department of New York (FDNY) have joined forces to shorten the response time for 1.7 million annual emergency calls.
The American University of Beirut in Lebanon will use machine learning to help farmers save water used for plant irrigation and food production.
Assessing 20 funded companies trying to find out how much cloud credits would be useful for each company, it became clear that their approaches were not always the best for their particular project because they missed out on ways to do computing on the device compared to the cloud.
The report strongly recommends creating partnerships between companies to reduce redundancies and pool resources. For example, more than 30 applications focused on the use of AI to identify and control agricultural pests.
“We believe that partnership between organizations with deep industry knowledge and organizations with technical expertise is the shortest term opportunity to develop and operationalize the use of AI for social purposes,” the report says. “Joint forums in which organizations interested in using AI for social purposes could share missions and needs with technical experts can help facilitate connections.
Applicants in the areas of crisis response, economic empowerment and gender equality and inclusion were more likely to face a lack of meaningful data sets, while those working in the health, environment, education, public and social sectors were more likely than others to have access to the necessary data.
Google’s review of the applications revealed a lack of experience in the use of AI systems. This is consistent with the report’s finding that 55% of applicants for non-profit and 40% of applicants for for-profit social enterprises had no previous experience with AI.
The report also often found a vague understanding of the limitations of AI and what it takes to bring a project from concept to completion.
“Many applicants who were new to the CI needed a better understanding of the nature of the data roles required, and others found it difficult to compete with private sector companies in hiring technical talent,” the report says. “During our review process, we found that even the oldest companies underestimated the time and resources required to prepare and maintain the data for the use of algorithms.
The lack of IT talent that affects many companies is particularly acute in the social impact world, but private companies that typically have more resources or internal IT teams also face challenges. An Autumn 2018 PricewaterhouseCoopers survey of executives in more than 60 countries revealed that only 4% say they have successfully implemented AI for their businesses.
The report does not mention any direct limitations on computing power, but a lack of computing power could reduce the type of model attempts that can lead to breakthroughs.
The Global AI Impact Challenge ends in February. Google has not yet determined whether a second challenge of $25 million will take place.