Industrial and Commercial Uses of Artificial Intelligence Michael

  • Slides: 42
Download presentation
Industrial and Commercial Uses of Artificial Intelligence Michael Brost

Industrial and Commercial Uses of Artificial Intelligence Michael Brost

Presentation Overview �History �Theory �Examples �Future

Presentation Overview �History �Theory �Examples �Future

The Inspirations

The Inspirations

The Turing Test �Man communicating with Man �Man communicating with Machine �http: //www. cleverbot.

The Turing Test �Man communicating with Man �Man communicating with Machine �http: //www. cleverbot. com/

John Mc. Carthy �Coined the term “Artificial Intelligence” �First Conference to discuss and debate

John Mc. Carthy �Coined the term “Artificial Intelligence” �First Conference to discuss and debate AI

The Power of the Mind versus the Power of the Machine

The Power of the Mind versus the Power of the Machine

CPU and Computer Performance �MIPS �SPEC �Linpack �FLOPS �STREAM

CPU and Computer Performance �MIPS �SPEC �Linpack �FLOPS �STREAM

MIPS �MIPS – Million Instructions Per Second �Older way of calculating Speed �Controversial

MIPS �MIPS – Million Instructions Per Second �Older way of calculating Speed �Controversial

Flops �Flops- Floating-Point Operations Per Second �(Cores x Clock x FLOPS)/Cycle � 1 Cycle

Flops �Flops- Floating-Point Operations Per Second �(Cores x Clock x FLOPS)/Cycle � 1 Cycle ≈ 4 Flops

Man or Machine? � Titan • • Cray Inc. 18, 688 nodes 4, 352

Man or Machine? � Titan • • Cray Inc. 18, 688 nodes 4, 352 sqft 17. 6 Petaflops

Man or Machine? �Unit • • • Multiples Kilo Mega Giga Tera Peta Exa

Man or Machine? �Unit • • • Multiples Kilo Mega Giga Tera Peta Exa

Man or Machine? � Titan • • • Cray Inc. 18, 688 nodes 4,

Man or Machine? � Titan • • • Cray Inc. 18, 688 nodes 4, 352 sqft 17. 6 Petaflops 2. 5 Million times faster! ~27 PFlops Potential

Man or Machine? � Titan • • • Cray Inc. 18, 688 nodes 4,

Man or Machine? � Titan • • • Cray Inc. 18, 688 nodes 4, 352 sqft 17. 6 Petaflops. 7 -1. 8% of a Human Mind � The Human Mind • 88. 5 cubic cm • ~100 Billion Neurons • 1 -2. 5 Exaflops (1, 000 - 2, 500 Petaflops)

Man or Machine? �Gordon Moore’s Law • “Number of transistors on integrated circuits would

Man or Machine? �Gordon Moore’s Law • “Number of transistors on integrated circuits would double every year for 10 years. ” –Moore • “Number of transistors on integrated circuits would double every year 18 months for 10 years. ” – Moore • “Computer performance would double every 18 months [due to increasing chip performance and technology]. ” – David House, 1975

Man or Machine?

Man or Machine?

The Chinese Room Argument Example

The Chinese Room Argument Example

The Chinese Room Argument �“Would the computer actually know what it was doing? ”

The Chinese Room Argument �“Would the computer actually know what it was doing? ”

Uses In the World Today

Uses In the World Today

Definition �The theory and development of computer systems able to perform tasks that normally

Definition �The theory and development of computer systems able to perform tasks that normally require human intelligence.

Watson: Jeopardy

Watson: Jeopardy

Watson: Jeopardy �Not hooked up to the internet �Downloaded many archives, encyclopedias and dictionaries

Watson: Jeopardy �Not hooked up to the internet �Downloaded many archives, encyclopedias and dictionaries to Watson �One Problem ….

Watson: Jeopardy �Language and context.

Watson: Jeopardy �Language and context.

Watson: Jeopardy �Solution? • “Evidence Profile”

Watson: Jeopardy �Solution? • “Evidence Profile”

Page. Rank

Page. Rank

Google’s Page Rank Algorithm � 1993 • First Web Search Engine • W 3

Google’s Page Rank Algorithm � 1993 • First Web Search Engine • W 3 Catalog • Highly inefficient

Google’s Page Rank Algorithm � 1996 • Back. Rub • Stanford Servers • Larry

Google’s Page Rank Algorithm � 1996 • Back. Rub • Stanford Servers • Larry Paige and Sergey Brin

Google’s Page Rank Algorithm � 1997 & 1998 • Group acquires initial capital to

Google’s Page Rank Algorithm � 1997 & 1998 • Group acquires initial capital to “Google Inc. ” �A play on words of “Googol” � 1 x 10 ^ 100 • Google breaks “Top 100 Web Site and Search Engines”

Google’s Page Rank Algorithm �Core Principles • “It’s best to do one thing really,

Google’s Page Rank Algorithm �Core Principles • “It’s best to do one thing really, really well. ” • “Focus on the user and all else will follow. ” • “You can make money without doing evil. ”

Google’s Page Rank Algorithm �Page • • Rank Location Time Existed Pages linked to

Google’s Page Rank Algorithm �Page • • Rank Location Time Existed Pages linked to the Website Page Rank Checker

Stocks & Wall Street

Stocks & Wall Street

A. I. in Wall Street �Rebellion Research • $7 million hedge fund • 6

A. I. in Wall Street �Rebellion Research • $7 million hedge fund • 6 employees �Mathematicians �Computer Science �Business Analysts

A. I. in Wall Street “It's pretty clear that human beings aren't improving, but

A. I. in Wall Street “It's pretty clear that human beings aren't improving, but computers and algorithms are only getting faster and more robust. ” Spencer Greenberg Rebellion Research CEO

A. I. in Wall Street �“Star” Software Program �Collects stock data overnight • 4,

A. I. in Wall Street �“Star” Software Program �Collects stock data overnight • 4, 000+ stocks • 70 different countries

A. I. in Wall Street � 150 points of view � 12 different strategies

A. I. in Wall Street � 150 points of view � 12 different strategies �$7 to $15 million

Artificial Intelligence into the Future

Artificial Intelligence into the Future

AI in the Future �Cleaning up the environment �Improved security systems �Advanced medicine �Diagnosing

AI in the Future �Cleaning up the environment �Improved security systems �Advanced medicine �Diagnosing conditions �….

�“It’s a bug that cars were invented before computers. ” Eric Schmidt Google Inc.

�“It’s a bug that cars were invented before computers. ” Eric Schmidt Google Inc.

The Driverless Car

The Driverless Car

The Driverless Car

The Driverless Car

The Driverless Car

The Driverless Car

The Driverless Car �Numbers • 90% �Reduce the number of accidents by 90% •

The Driverless Car �Numbers • 90% �Reduce the number of accidents by 90% • 90% �Reduce number of cars by 90% • 405 Billion �Save $405 Billion in revenue • 3 � 3 States have passed laws • 300, 000 � 300, 000 miles have been logged

Questions

Questions