Page 18 - A U4SSC deliverable - Accelerating city transformation using frontier technologies

and 40 per cent of the total waste generated worldwide is not disposed of properly, thereby significantly
contributing to the increase in health and environmental risks, particularly in developing countries
where most waste is likely to end up in landfills or openly burned.
25

The effective management of waste becomes a significant challenge for cities to tackle, and AI is
already making great strides in alleviating the pressure cities are facing in this area. AI-powered sorting
technology can modernise the waste management process by improving the efficiency and productivity
of waste management. Increasingly, AI and machine learning are helping to sort waste in municipalities.
Autonomous robots are being trained using images of different types of trash in order to identify them,
and to, eventually, be able to sort them accordingly for better management.

AI also covers the training of robots to recognise recyclable materials, from plastic bottles to other
containers in order to be able to sort them into recycling bins or to recover these materials. These
robots can be twice as effective as humans and can significantly improve the recycling rate. Intelligent
26
trash bins fitted with vison sensors are being trained by AI to identify the type of trash that is being
thrown away and sort them into bins by types. These trash bins can also detect when the bin is full,
thereby optimising the route for collection trucks. In the United States, sorting robots are being
27
deployed in Sarasota, Florida. They are able to sort 70 to 80 items per minute, twice as fast as humans
and with greater accuracy.
28

“AI and machine learning are helping to sort waste in municipalities.
Autonomous robots are being trained using images of different
garbage in order to identify their types and sort them accordingly”

AI also plays a prominent role in improving health care in cities. The total investment of AI in health
care is expected to reach USD 6.6 billion by 2021. It is estimated that the use of AI applications in the
health sector could achieve an annual saving of USD 150 billion by 2026, from virtual nursing assistants
and administrative workflow assistance to automated image diagnosis and robot-assisted surgery. 29

Medical diagnosis accounts for almost a third of all AI applications in health care. Error in diagnosis
accounts for approximately 10 per cent of patient deaths and up to source 17 per cent of hospital
complications. AI can help overcome this challenge by reviewing vast amounts of health data and
medical records, generating powerful algorithms that support medical personnel to diagnose patients
correctly.

The ability to diagnose a patient correctly will become more important as the world’s population
continues to age and life expectancy continues to rise at the same time. The application of AI in
medical diagnosis can come in several forms. For example, chatbots can facilitate virtual nursing
30
assistants by using speech recognition to identify patient symptoms, form a simple diagnosis and
provide recommendations by comparing the reported symptoms against a database of disease. Virtual
nursing can also reduce unnecessary hospital visits and lessen the burden on medical professionals. It
is estimated that an AI-powered virtual nursing assistant can save up to USD 20 billion annually. The
31

8 Accelerating city transformation using frontier technologies | A U4SSC deliverable