Artificial Intelligence for Social Impact February 8, 2018 Dr. Cara LaPointe Senior Fellow Georgetown University
What is Artificial Intelligence? 2
Artificial Intelligence: A Working Definition The capability of a machine to imitate intelligent human behavior. Miriam Webster Dictionary 3
The Father of Artificial Intelligence Alan Turing: 1935 The Universal Turing Machine: A precursor to modern computers 1950 The Turing Test: Could a machine fool a human interrogator? He believed in the possibility of computing machines: (1) Learning from experience (2) Solving problems by means of searching through the space of possible solutions, guided by rule-ofthumb principles heuristic search http://www.alanturing.net/turing_archive/pages/reference%20articles/what_is_ai/what%20is%20ai05.html 4
What does that really mean? 5
Robotics Machines performing physical action Artificial Intelligence Machines performing cognitive action 6
A Way of Looking at Cognitive Action: The OODA Loop Observe Orient Decide Act 7
The OODA Loop Artificial Intelligence Observe Orient Decide Act 8
The OODA Loop Autonomous Systems Observe Orient Decide Act 9
Understanding and Applying AI and Autonomy 10
How does it work? 11
The Functions of Intelligence Learning Reasoning Problem-solving Perception Language-understanding 12
https://machinelearningmastery.leadpages.co/leadbox/147764973f72a2%3a164f8be4f346dc/5752754626625536/
Machine Learning A Simple Example 14 https://medium.com/@ageitgey/machine-learning-is-fun-part-3-deep-learning-and-convolutional-neural-networks-f40359318721
Deep Learning https://nivdul.files.wordpress.com/2015/11/nivdul_deep_learning.png?w=700&h=367 15
What are the promise and challenges in applying AI? 16
Ai is transformative Consumer Health Finance Retail Government Energy Transport Industrial Other Smart Assistants Chatbots Search Personalization Augmented Reality Robots Enhanced Diagnostics Drug Discovery Patient Care Research Sensory Aids Algorithmic Trading Fraud Detection Research Personal Finance Risk Mitigation Support Experience Marketing Merchandising Loyalty Supply Chain Security Defense Data Insights Safety & Security Resident Engagement Smarter Cities Oil & Gas Exploration Smart Grid Operational Improvement Conservation Autonomous Cars Automated Trucking Aerospace Shipping Search & Rescue Factory Automation Predictive Maintenance Precision Agriculture Field Automation Advertising Education Gaming Professional & IT Services Telco/Media Sports Source: Intel forecast Source: Tang, Intel, 2017
What are the challenges of AI? INPUTS OUTPUTS BLACK BOX 18
AI algorithms are not always transparent, so take the time to understand the algorithm and the data. Understand that algorithms are biased. Recognize this and act accordingly. 19
Recognizing correlation is not the same as either understanding or identifying causation. 20
Take the time to understand the technology, including its limitations, and then thoughtfully design and apply it. 21
How can AI and autonomous systems be leveraged for social impact? 22
WeRobotics and the Open AI Challenge 2018 Uses machine learning and computer vision to automatically identify features of interest in aerial imagery. WeRobotics has launched a competition to create Imagery of South Pacific Island. This will help accelerate and improve humanitarian and development efforts across the South Pacific. https://werobotics.org/blog/2018/01/10/open-ai-challenge/ 23
TraffickCam: Fighting Human Trafficking with AI The TraffickCam app was launched to help combat sex trafficking by uploading photos of the hotel rooms they stay in. The purpose of TraffickCam is to create a database of hotel room images that an investigator can efficiently search, in order to find other images that were taken in the same location as an image that is part of an investigation. https://traffickcam.com/about TraffickCam was launched in June 2016 and more that 56,000 people sign up in the first two weeks. https://www.snopes.com/2016/07/03/takingpictures-of-your-hotel-room-could-help-stophuman-trafficking/ 24
What are the key takeaways? 25
Key Takeaways on AI and Autonomy for Social Impact There is a lot of misinformation about AI and autonomous systems They are technology tools that, like all technology, should be designed thoughtfully to mitigate any potential negative consequences and to take advantage of their potential for positive social impact AI can take massive amounts of data and learn how to turn it into useable information; autonomous systems are machines that take action based on useable information Take the time to understand what the information means and how it is useable; understand the limitations and the applicability 26
Artificial intelligence (AI) and autonomous systems are powerful tools to amplify social impact efforts if they are thoughtfully developed and applied. 27
Questions? Thoughts? Email: Cara.LaPointe@georgetown.edu
[BACKUP] 29
The key is understanding what the AI algorithm is learning, and then interpreting and applying that information appropriately. 30
31