Dr. Usama FayyadChairman & CEO - Open Insights, Co-Founder & Advisory CTO –– OODA Health, Inc.
Making AI work in practice: Why Data is so Essential to AI and Towards Data Science Standards
Artificial Intelligence (AI) has been receiving a lot of hype as a magical solution for many difficult problems. Some have started to worry that AI would not only take over their jobs but also take over control of work and perhaps personal life. In the meantime, many businesses feel pressured that if they are not leveraging AI now, they may miss the boat on the next wave to digitization and business automation.
I have been invited by many executive teams and corporate boards, both public and private, to explain what parts of this technology are real, what parts are hype, and most importantly what they need to do in their business in the short and medium terms to make sure they are keeping up with the best technology practices globally.
In this talk, I attempt to demystify most of these concerns by simply describing in reality what has worked and what has not. The main theme is that the technology can be extremely useful and powerful, but is nowhere near the hype associated with it. The main themes I will cover are:
- A brief history of AI: the fundamental challenges and why we are, and will be safe, from replacement by robots in the foreseeable future… This will help us to crystallize what has actually worked and why
- Winter is coming: We are on the verge of a third AI winter. But this does not mean that we should ignore the parts of the technology that works.
- It turns out, while most of AI has failed (with some very notable exceptions). Machine learning (ML) is behind most of what has worked. This has major implications as most of ML algorithms succeeded not because of great algorithm design, but primarily because a lot more data became available.
- Data and Data Science are critical to making most of AI work. We thus go into what the issues are in making data, especially BigData – which allows for the proper utilization of the majority of data in any organization (unstructured data).
- Despite all the hype, with the right Data in place, ML/AI are making their way into all kinds of business operations today as more companies explore how they can put their data to work. Among the most common reasons to apply ML to business processes is the ability of the technology to perform certain functions at scale, which would otherwise require significant amounts of time and resources. Data is the fuel that powers a machine learning solution and machine learning is the key to make most AI algorithms practical.
- Licensed to Analyze? : The industry of Data Science is thus becoming super critical, yet we struggle to define roles, standards and an ability to scale and assess capabilities, We cover our industry initiative (IADSS) to bring standards and understanding to this space
Usama founded Open Insights as a technology and consulting firm in 2008 after leaving Yahoo! to enable enterprises to get value out of their data assets and optimize or create new business models based on the new evolving economy of interactions. Leveraging both technology and strategic consulting Open Insights deploys data-driven solutions to grow revenue from Data assets through BigData strategy, new business models on data assets, and deploying data science, AI/ML solutions. Usama is also Co-Founder & CTO at OODA Health, Inc a VC-funded company aiming to liberate the healthcare system from administrative waste by leveraging AI/automation to create real-time/retail-like experience in payments in healthcare.
From 2013-2016 Usama served as Global Chief Data Officer & Group Managing Director at Barclays Bank in London, after launching the largest tech startup accelerator in MENA following his appointment as Executive Chairman of Oasis500 in Jordan by King Abdullah II in 2010. His background includes Chairman and CEO roles at several startups, including Blue Kangaroo Corp, DMX Group and digiMine Inc. He was the first person to hold the Chief Data Officer title when Yahoo! acquired his second startup in 2004. At Yahoo! he built the Strategic Data Solutions group and founded Yahoo! Research Labs where much of the early work on BigData made it to open source and led to Hadoop and other open source contributions. He has held leadership roles at Microsoft (1996-2000) and founded the Machine Learning Systems group at NASA’s Jet Propulsion Laboratory (1989-2005), where his work on machine learning resulted in the top Excellence in Research award from Caltech, and a U.S. Government medal from NASA.
Usama has published over 100 technical articles on data mining, data science, AI/ML, and databases. He holds over 30 patents and is a Fellow of both the AAAI and the ACM. Usama earned his Ph.D. in Engineering in AI/Machine Learning from the University of Michigan. Ann Arbor. He has edited two influential books on data mining/data science and served as editor-in-chief on two key industry journals. He also served on the boards/advisory boards of private and public companies including: Criteo, Invensense, Exelate, RapidMiner, Stella.AI, Virsec, Silniva, Abe.AI, NetSeer, Choicestream, Medio, and others. He is on advisory boards of the Data Science Institute at Imperial College, AAI at UTS, and The University of Michigan College of Engineering National Advisory Board. He serves on the Board Advisory Committee to Nationwide Building Society in the UK and on the Advisory board of the WEF Global Center for Cybersecurity. He is an active angel investor and advisor in many early-stage tech startups across the U.S., Europe and the Middle East.
Prof. Ankur M. TeredesaiFounder & CTO – KenSci, Professor – University of Washington Tacoma
AI for Health: How to Make the World a Better Place with Machine Learning
The world is struggling with the COVID-19 pandemic of epic proportions that was in no way predictable. This has brought increased attention to the field of healthcare and role that AI can play in helping predict and improve outcomes at global scale. Pandemics are rare but it is important to note that globally three in five deaths are actually attributed to just four major non-communicable diseases: Cardiovascular disease, cancer, chronic lung disease and diabetes. Yet we are far from effective care management and AI driven assistive intelligence that can be widely deployed to help improve patient outcomes. Why is AI/machine learning in healthcare settings so tough to deploy? Is it the accuracy of our ML models? Is it the complexity of the hypothesis spaces? It is the lack of right data perhaps, or wide availability of scalable infrastructure? Or are we just completely missing the proper governance frameworks to ensure compliance for AI solutions in healthcare? In this talk, I present both successes and failures of my decade long journey in Healthcare AI of building and deploying machine learning solutions across the globe from Seattle, to Singapore and Scotland. I begin by showing how simple but accurate ML models are necessary but not sufficient and how artificial intelligence can be more assistive when we include fairness, accountability and transparency in our modeling frameworks at both N=n and N=1 levels. Then, I briefly overview several new approaches for enhancing interpretability and draw attention to latest efforts in modeling the math behind pandemics drawing from prior lessons. I conclude by drawing inspiration from early pioneers in computing and why I believe AI can not only just make the world a better place, but actually save lives.
Prof. Ankur M. Teredesai is a computer scientist and Healthcare AI expert. He is the founder and Chief Technology Officer of KenSci, the global leader in Healthcare AI. Ankur holds a tenured full professorship at the School of Engineering & Technology, University of Washington, at Tacoma Washington. Prof. Teredesai has published over 100 papers in machine learning and also successfully implemented numerous AI driven applications that are in use in various industries. Since 2009, his primary focus is to make AI assistive for healthcare. From advancing our understanding of population health and utilization for multimorbid conditions to implementing Responsible AI techniques for personalized precision health, Ankur Teredesai has been instrumental in helping healthcare organizations go through digital transformations using AI. In May 2015, after years of collaborative and applied research on large clinical and claims datasets, Prof. Teredesai founded KenSci, a spin-out form the University of Washington.
Ankur is an advocate for diversity and inclusion, women in computing and enabling non-traditional students to consider computing careers. In 2019, he led as the General Co-Chair the prestigious ACM KDD (Knowledge Discovery and Data Mining) conference. When not designing healthcare AI solutions, he loves hiking up to remote villages in the high Himalayas.