Human + Machine How AI is Radically Transforming and Augmenting Lives and Businesses Are You Ready? Xavier Anglada Managing Director Accenture Digital Lead in MENA and Turkey @xavianglada TM Forum 1
Meet SpotMini, your new teammate A nimble robot that handles objects, climbs stairs, and will operate in offices, homes and outdoors BostonDynamics TM Forum 2 2
Hey buddy, can you give me a hand? TM Forum 3 3
How about his brother, Spot? TM Forum 4 4
AI IS CREATING THE SINGLE BIGGEST TECHNOLOGY REVOLUTION THE WORLD HAS EVER SEEN Mainframe Client-Server and PCs Web 1.0 ecommerce AI Web 2.0, Cloud, Mobile Big Data, Analytics, Visualization IoT and Smart Machines Quantum Artificial Intelligence Quantum Computing 1950 Turing Test 5
AND IS EVOLVING AT EXPONENTIAL PACE TM Forum 6 6
IMPLEMENTING AI CAN BE DONE TODAY CURRENT COMBINATORIAL TECHNOLOGIES ARE MAKING AI AFFORDABLE, DOABLE, AND AVAILABLE TODAY. DEMOCRATIZATION OF AI SKILLS DATA TO TRAIN AND FUEL AI IS MORE ACCESSIBLE TODAY HIGH PERFORMANCE COMPUTING POWER IS MORE READILY AVAILABLE PROLIFERATION OF AI PRODUCTS & SOLUTIONS 7
AND THE TIME TO ACT IS NOW LOW BARRIERS TO AI SIGNIFIES THAT NOT ONLY CAN YOUR ENTERPRISE REALIZE THE BENEFITS, BUT COMPETITORS WILL AS WELL. LOW BARRIERS TO ENTRY WILL ONLY ACCELERATE THE RATE OF COMPETITION. EXPONENTIAL GROWTH ACCELERATES HIGH PERFORMING ORGANIZATIONS AND WILL LEAVE THE SLOW MOVERS BEHIND. WHY IS AI DIFFERENT THAN OTHER TECH TRENDS? Combinatorial technologies have fueled the exponential growth of AI being a fast follower is not enough Autonomous self learning and a low cost to scale accelerates the diffusion of AI across businesses and society. Is a game-changer to society; causing dramatic benefits and creating new challenges. 8
IF YOU WERE FULLY ACCOUNTABLE FOR THE DECISION OF AN AI ALGORITHM, WOULD YOU BLINDLY TRUST IT? From To TRUST IS THE KEYWORD TM Forum 9 9
AI: FROM BLACK BOX TO GLASS BOX Black box AI works well Simple pattern recognition tasks (low cost of failure) Closed loop systems (need for real-time response, i.e. Amazon real-time pricing) Interactive response systems (i.e. chatbots) Existing models and their interpretation Glass box AI is required Business situations where an explanation of the underlying reasoning is critical for decision-makers 10
EXAMPLES OF POOR AI DECISION FROM 2018 CHINESE BILLIONAIRE S FACE IDENTIFIED AS JAYWALKER Traffic police in major Chinese cities are using AI to address jaywalking. The AI system in the southern port city of Ningbo however recently embarrassed itself when it falsely recognized a photo of Chinese billionaire Mingzhu Dong on an ad on a passing bus as a jaywalker. AI WORLD CUP 2019 PREDICTIONS ALMOST ALL WRONG AI researchers at Goldman Sachs, German Technische University of Dortmund, Electronic Arts, etc. ran machine learning models to predict outcomes for the multi-stage competition. Most however were totally wrong, with only EA which ran its simulations using new ratings for its video game FIFA 18 correctly favoring winner France. Source: Medium.com / 2019 in Review: 10 AI Failures UBER SELF-DRIVING CAR KILLS A PEDESTRIAN The Uber vehicle was in autonomous mode, with a human safety driver at the wheel. Uber discovered that its selfdriving software decided not to take any actions after the car s sensors detected the pedestrian. IBM WATSON COMES UP SHORT IN HEALTHCARE Watson AI Health has not impressed doctors. Some complained it gave wrong recommendations on cancer treatments that could cause severe and even fatal consequences. AMAZON AI RECRUITING TOOL IS GENDER BIASED Amazon HR reportedly used an AI-enabled recruiting software between 2014 and 2017 to help review resumes and make recommendations. The software was more favorable to male applicants because its model was trained on resumes submitted to Amazon over the past decade, when many more male candidates were hired. 11
UNDERSTANDING MACHINES: EXPLAINABLE AI (XAI) If AI lacks the ability to explain itself, then the risks of making a wrong decision may outweigh the benefits it can bring in terms of the speed, accuracy and efficiency 12
WHY DO WE NEED XAI? COMPLIANCE TRANSPARENCY ETHICS Provides with an auditable record, including all factors and objects associated with the decision-making Eliminates concern that the organization is hiding or doesn t even know how a machine is affecting an outcome for a critical decision Ensures that there is an auditable and provable way to defend algorithmic decisions as being fair and ethical 13
XAI, THE FUTURE FOR THE INTELLIGENT ENTERPRISE 1. Explainable and more responsible AI will be the backbone of the intelligent systems of the future that enable the intelligent enterprise 2. XAI is ready to work alongside its human counterparts. Businesses can create a collaborative, powerful new member of the workforce by raising AI for responsibility, fairness and transparency 3. In playing this role, Explainable AI won t replace people, but will complement and support them so they can make better, faster, more accurate and more consistent decisions 14
EXPLAINABLE AI SYSTEMS WILL PLAY A PIVOTAL ROLE THROUGH THEIR ABILITY TO 15
EXAMPLES OF EXPLAINABLE AI MARKETING CAMPAIGNS Increase relevancy of messaging, offers and creative to match the real and most important interests of predicted buyers ONLINE RECOMMENDATIONS Predicts what a customer wants and why for more precise and auditable, personalized recommendations FRAUD DETECTION Predict and explain why certain transactions are fraudulent ANOMALY DETECTION Identify and explain anomalous behavior, events, equipment failure or device activity for fast, preventative action CREDIT & LOAN DECISION-MAKING Auditable, transparent predictive models for more accurate, explainable and compliant credit and decision making 16
SO-WHAT: REDEFINING MATHS HUMAN + MACHINE 1 + 1 > 2 17
THANK-YOU