NISO Plus: Artificial Intelligence and Machine Learning 101
Download as PPTX, PDF1 like198 views
The document discusses the rapid increase in data generation and the limitations of the current information economy, highlighting that traditional methods of data analysis are not scalable. It outlines the evolution of human innovation related to artificial intelligence, differentiating between various AI concepts and emphasizing the current focus on narrow AI and machine learning. Additionally, it addresses what AI/ML technologies can and cannot do, particularly their capabilities in handling large data volumes and the importance of human involvement.
NISO Plus: Artificial Intelligence and Machine Learning 101
- 1. NISO PLUS 2020 Artificial Intelligence & Machine Learning 101 Jason Chabak February 24th, 2020
- 3. Strictly confidential, not for distribution without the prior consent of Yewno DATA VOLUMES ARE EXPLODING ● More data has been created in the past two years than in the entire previous history of the human race ● By 2020, about 1.7 megabytes of new information will be created every second for every human being on the planet ● By 2025, our accumulated digital universe of data to around 160 zettabytes, or 160 trillion gigabytes Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program) Annual Size of the Global Datasphere Source: IDC’s Data Age 2025 study, sponsored by Seagate, April 2017 One zettabyte is equal to one sextillion bytes or 1021 (1,000,000,000,000,000,000,000) bytes or, one zettabyte is roughly equal to a trillion gigabytes.
- 4. The traditional information economy is not scalable. “It would take over 3 trillion years for a team of 10,000 analysts to read all of the unstructured information currently available today.” Information is fragmented. Information is not the same as knowledge.
- 5. HOW CAN WE DEAL WITH IT?
- 6. Strictly confidential, not for distribution without the prior consent of Yewno A STORY OF HUMAN INNOVATION Artificial intelligence stands in a long line of human innovation in the creation, recording, storing, codifying, and dissemination of cultural knowledge ● 3500 BC: Written language ● 1790 BC: Code of Hammurabi ● 500 BC: Egyptian papyrus rolls (scrolls) ● 300s BC: Library of Alexandria ● 100 AD: Codices ● 1791: The French cataloging code ● 1950s: Natural Language Processing ● 1960: Full-text search, hypertext ● 1963: ASCII code ● 1968: MARC records ● 1980s: Neural networks ● 1989: The Web ● Etc., etc., etc…. Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program) Linear B tablet Roman codex Dead Sea scroll ASCII code Library of Alexandria Computational linguistics + machine learning + graph theory
- 7. WHAT ARE WE TALKING ABOUT?
- 8. Strictly confidential, not for distribution without the prior consent of Yewno DIFFERENTIATING THE CONCEPTS • Artificial Intelligence • Narrow AI • General AI • Machine Learning • Deep Learning • Artificial Neural Networks • ... • ... • Artificial Super-Intelligence Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program) Source: https://towardsdatascience.com/cousins-of-artificial-intelligence-dda4edc27b55
- 9. Strictly confidential, not for distribution without the prior consent of Yewno WHERE ARE WE? • Where are we in the pursuit of AGI? • C. Elegans (1-millimeter flatworm) brain replication • Successfully simulated in software and inserted into Simple Lego robot • 302 Neurons replicated • Human brain estimated to contain 100,000,000,000 neurons • Computing capacity needs to be exponentially increased • We’ve still got a ways to go, so for today, we’re focusing on Narrow AI & ML Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program) Source: Cornelia I. Bargmann, Rockefeller University; WormAtlas; Nature
- 10. WHAT AI/ML CAN’T DO
- 11. Strictly confidential, not for distribution without the prior consent of Yewno CURRENT AI/ML/DL UN-REALITIES • 100% accuracy • 100% unbiased • Operate without human intervention • Have no adverse effects • Scale at current technological infrastructure • Take ALL of our jobs • Take over the world….yet Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program)
- 12. WHAT AI/ML CAN DO
- 13. Strictly confidential, not for distribution without the prior consent of Yewno THE CURRENT LANDSCAPE OF NARROW AI & ML • Ability to perform repetitive, mundane tasks – and improve over time • Intelligent Augmentation of current workflows • Reduction – not elimination - of errors • Ingest and analyze orders of magnitude more data than any human – or group of humans – would be able to • Quality out only as good as the quality in • Human element is of utmost importance Jaguar (animal) Jaguar (automobile) Jaguar (beverage) Jaguar (tank) Jaguar (software) Jaguar (athletic program)
- 14. THANKS!