Thesis Topic: AI for Sustainable Development

[Audio] Hi, My name is Monica Uttarwar, and I am a new PhD student at Faculty of Science and Technology. I would like to introduce my thesis topic for 3 minute presentation today. I am planning to work on AI for Sustainable development..

[Audio] AI or Artificial Intelligence could be a key enabler for achieving UN's sustainable development goals. However, there is a need to understand the power and limitations of AI, and how we can harness it to our benefit, by understanding the power and limitations of the underlying computational engine known as the Machine learning core and enhancing its power with novel algorithmic models. The focus of my PhD work would be to develop an innovative algorithmic framework, to apply AI . technologies for achieving some of the key SDG goals..

[Audio] Recently, there is increase in use of AI in several application areas, such as: Robotics and Automation, Self Driving Cars, Healthcare sector, Energy and Environment, Telecommunications and Transportation. However, to develop AI based technologies for extreme environments and complex application settings, such as achieving some of the key UN's SDG goals, it is important to understand the current capabilities , as well the shortcomings of AI technologies, and how it can be applied for solving complex problems which human brains can easily solve..

[Audio] The underlying computational engine behind AI using different types of machine learning models as depicted in this screen-shot. But they are still not able to achieve the capabilities equivalent power of a human brain, and work under complex application settings, such as UN SDG scenarios. Perhaps by combining and enhancing some of these algorithm models, it could be targeted for building technology platforms for achieving SDG goals. But these have not been done yet, and I plan to focus on these for my research..

[Audio] Let us look at the power and limitations of current AI technologies. At present, we could just tap into i-level (individual level) intelligence of human brains and built AI technologies to achieve this level of automation. In fact, there is more we do NOT know about the brain, than what we do know. With i-level (individual level) intelligence, we fully understand low-level functions and make the computers intelligent enough to solve small problems. This includes localized and limited sensory capabilities of sight, hearing, smell, taste and touch, few motor functions, such as object grasping and limited sensory-motor functions such as obstacle avoidance. The higher-level functions of brain at i-level are still not fully understood, hence could not be transformed into machine intelligence or AI. These higher-level functions of brain, include perception, situational awareness, decision making under uncertainty, learning and adaptation..

[Audio] The next level is the g-level ( group-level) intelligence, and again, AI could be developed to some extent to perform the low-level functions at group level(g-level), such as making sense of co-operative data gathering, information exchange, co-operative object detection and co-operative manipulation. Some of the high-level functions include shared situational awareness, consensus finding, co-operative decision making, group formation, multi-agent learning. So, AI technologies suffer significantly in interpreting collective behaviours, as collective behaviours are not just sum of individual behaviours, but they emerge at the society level..

[Audio] To summarize and make sense out of the hype surrounding the capabilities of AI, on what it can do and what it can't, we can categorize AI capabilities into narrow AI (the current level), which is at i-level, and not as powerful as human brain, but is superior to mouse brain. The next goal in evolution of AI is AGI ( Artificial General Intelligence), with i-level and some g-level capabilities, and it will be then superior to human brain. We are yet to achieve this goal, the AGI goal, and then the final goal, which will comprise full i-level and g-level capabilities to be achieved, called as the ASI ( Artificial Super Intelligence), and it would be a super-brain, with power equivalent to several combined human brains..

[Audio] To be able to develop AI technologies for being enabling technologies and for achieving some of the SDG goals, the current machine learning engines and associated, underlying algorithmic frameworks enabling the intelligence levels are far from what is required. So, my PhD focus will be to develop a set of innovative algorithmic models for building AI enabled platforms for addressing some of the key SDG goals..

[Audio] Thank you for listening, and if anyone has any questions, please email me at this email id, and I would be happy to answer any questions. Thank you.