” Social Impacts of Machine Intelligence”
Significant media hype has stoked fears that A.I. will be harmful to our world. Many articles hypothesize that significant job losses will result from this technology. The Future of Life Institute has penned letters that attempt to control the proliferation of A.I.-based weapon systems, and to ensure that research into the ethics and control of A.I. occurs. We hear talk about the singularity, a moment when machine intelligence will exceed human intelligence growing exponentially beyond our control. There is no consensus on many of these issues, or even on a clear definition of A.I.
The panel consisting of experts in a variety of machine intelligence fields will shed light on these topics and will also answer questions from the audience.
Key discussion areas will include:
- What should the definition of artificial intelligence be?
- Potential job losses from A.I: real or not?
- Should we be afraid of A.I. systems?
- What are some of the legal and ethical implications of implementing A.I. systems?
Moderator: Gary Saarenvirta
Development and adoption of artificial intelligence (AI) technology in products and services is exploding and careful assessment of the legal risks of doing business in the new AI-powered marketplace is important. Isi Caulder, partner, lawyer and a registered patent and trademark agent with Bereskin & Parr LLP will discuss strategies for protecting intellectual property for AI related inventions and how to identify and mitigate risks and liabilities flowing from AI.
Isi Caulder is a partner, lawyer and a registered patent and trademark agent with Bereskin & Parr LLP. She is a member of the firm’s Electrical & Computer Technology practice group and co-leader of the firm’s Artificial Intelligence (AI) practice group. Isi guides technology companies developing and working with AI through the process of strategically capturing, managing, and monetizing their patent, design and other intellectual property rights in North America and worldwide.
“Using simulation to design, train and test automated vehicles (AVs)”
Richard Romano (EngSci 8T8, UTIAS MASc 9T0), Chair in Driving Simulation, University of Leeds
Automated vehicles (AVs) have been under development for over 50 years. In this time there have been advancements to sensors, software and computers, but incredible challenges still remain. Sensors need significant improvement to provide a level of capability similar to the human eye; but, most importantly, the software must be enhanced to meet the challenge. Based on recent crashes, it is obvious that further refinement is required to sense and react to objects and vehicles in the world; but this is only the tip of the iceberg. AVs need to interact in a safe and consistent method that maintains their priority and doesn’t lead to chaos on the roads! Simulation is an important tool to support the design and testing of AVs. It provides a safe environment for a variety of open and closed loop scenarios. However, testing complex closed loop scenarios where AVs interact with traditional traffic participants requires further Artificial Intelligence to replicate all the ‘players’. This sets the stage for many more years of AI research.
Speaker Bio: Prof. Richard Romano is the Chair of Driving Simulation at the University of Leeds, and has over 25 years of experience in academia and industry developing and testing automated vehicles. He started his career in 1990 as a research engineer at the University of Iowa, leading the development of the Iowa Driving Simulator and participating in seminal research in driving simulation, including simulator studies for the FHWA-funded Automated Highway System. In 1995 he joined ITT Automotive (now Continental Teves) where he led the brake simulation group, including the development of stability control (SCS) technologies. In 1997, he founded his own driving simulator company, Realtime Technologies, Inc (RTI), and built the most successful research driving simulator company in the US with multiple sales into Silicon Valley and many of the top automotive companies. He was recognized with a Tibbetts Award from the Small Business Administration (SBA) for the successful commercialization of a variety of driving simulation products with over $50 million in total sales. With the rapid deployment of automated vehicles, Richard returned to academia in 2015, joining the University of Leeds, where he is the lead or co-investigator on over £6 million of research activity while leading the development of the University of Leeds Driving Simulator. Prof. Romano holds a BASc in Engineering Science (8T8), an MASc from UTIAS (9T0), and a PhD in Human Factors (Iowa).
“Machine Intelligence: an overview of Machine Learning methods and applications”
Lorne Rothman, PhD, PStat, Principal, Data Sciences, SAS Canada
Machine Learning algorithms provide the brains beneath Machine Intelligence applications. They help to increase speed, automation and performance, while reducing the reliance on human input in the analytics process. We will briefly cover definitions of Machine Intelligence (or AI), and Machine Learning. We will present an overview of some common contemporary supervised, and unsupervised methods (e.g. Gradient Boosting, Support Vector Machines, Factorization Machines, Neural Networks & Deep Learning; Clustering, Support Vector Data Description) as well as their applications.
Speaker Bio: Dr. Lorne Rothman is a Principal Data Scientist at SAS Institute Canada, with expertise in the applications of Machine Learning, Text Analytics, Forecasting & Classical Statistics to solve real world problems. He has worked with a wide variety of private and public sector organizations across North America since joining SAS in 1998. He holds Bachelors and Masters of Science degrees from the University of Toronto and a Ph.D. in Ecology from the University of British Columbia, and is an accredited Professional Statistician (P.Stat.) with the Statistical Society of Canada.
The SAS Institute is the world leader in analytics and has customers in 148 countries with more than 83,000 software installations and 14,000 employees. SAS has been applying analytics to the toughest business problems for decades. SAS envisions a world where everyone can make better decisions, grounded in trusted data and assisted by the power and scale of SAS® Analytics. When decisions happen at just the right moment, advancements are set in motion and the world moves forward.
Reinforcement learning is a branch of machine learning or artificial intelligence that is loosely based on how humans and animals learn through interaction with their environment. As children we used random trial and error to take actions and observe the cause and effect. We stored our experiences and through repetition reinforced the cause and effect learning resulting in memories allowing us to recall what actions to take in similar situations. All through our lives we continue to learn in this manner. Saarenvirta’s bootcamp tutorial will define the differences between supervised learning and reinforcement learning. Several toy examples will be presented to illustrate the mathematics behind reinforcement learning. Different types of reinforcement learning will be described including model-based learning and several types of model free learning. The tutorial will also present how Saarenvirta’s company, Daisy Intelligence, uses simulation based reinforcement learning to assist retailers to make smarter merchandising decisions, including what products to promote each week, what prices to charge for each product at each store and how much inventory of each product to allocate to each store and distribution centre. Daisy’s methods are presented through real-world client examples and financial results are shared. The tutorial ends with a future vision of the autonomous enterprise and how the engineering profession will lead the development of practical artificial intelligence applications.
Speaker Bio: Gary Saarenvirta is one of North America’s preeminent experts in artificial intelligence having over 25 years’ experience working with leading global corporations to deliver revenue and profit growth. He founded Daisy Intelligence in 2003 bringing autonomous machine intelligence to clients in retail, insurance and healthcare. Headquartered in the Greater Toronto Area, the company operates an applied artificial intelligence (A.I.) software-as-a-service (SaaS) business delivering operational corporate decisions that are too complex for humans to make, resulting in efficiencies and profitability gains.
Currently, Daisy is revolutionizing optimization of merchandise planning for high volume retailers and fraud detection/risk management for insurance and banking. Using proprietary mathematical solutions and Daisy’s reinforcement learning based A.I. simulation platform, the company analyzes trillions of the trade-offs inherent in any complex business question and provides timely, actionable decision recommendations to help corporate clients grow total sales, improve margins and reduce fraud.
Gary is the former head of IBM Canada’s data mining and data warehousing practices. He was also at the helm of Loyalty Consulting Group, providing analytical services for one of the world’s most successful coalition loyalty programs, the AIR MILES® Reward Program. Gary holds both a BASc in Engineering Science (8T8) and MASc in aerospace engineering (0T2) from the University of Toronto.
Abstract: Matt Zeiler, a University of Toronto grad and world-renowned AI and machine learning expert, will explain how to build a world-changing company by taking concepts from research and academia to commercialization. After studying under AI pioneers Geoff Hinton and Jeff Dean, and receiving job offers from Google and Facebook, Matt returned to NYC to found Clarifai, an AI and visual recognition company that works with OpenTable, West Elm, Trivago, and others. He’ll share insights from his journey: merging business knowledge into his tech background, engineering hacks for developers, and pivotal moments in his career that helped him build an award-winning company.
Speaker Bio: Matthew Zeiler is the Founder and CEO of Clarifai, which uses machine learning and deep neural networks to understand images and launched with the world’s best published object recognition system. Prior to founding Clarifai, he interned at the Google Brain group where he learned how to use machine learning to tackle real world problems. While an undergraduate student at the University of Toronto, he founded Review-Mate, an online platform that sells course-specific review booklets to university students across North America. He holds a BASc in Engineering Science from the University of Toronto and a PhD in Computer Science from New York University.