On Chasing the Ether of Intelligent Systems Dr

Скачать презентацию On Chasing the Ether of Intelligent Systems Dr

16324c101cf725ccde1399cb8bd95987.ppt

Количество слайдов: 25

On Chasing the Ether of Intelligent Systems Dr. James Guffey james. a. guffey@boeing. com – (314) 232 -5042 25 Oct 2006 INCOSE Meeting – St. Louis MSG 06 -081332 -001. ppt

Name: Dr. James A. Guffey Senior Principal Engineer Key Technical Field: Technology Integration; Autonomous Technologies Phone: (314) 232 -5042 E-mail: james. a. guffeyt@boeing. com Dr. Guffey has been at Boeing/Mc. Donnell Douglas for 2 dozen and 3 years and has been heavily involved in the definition, implementation and management of a large number of avionics research and development programs for the Army, Air Force, Navy and DARPA, covering a wide variety of technology areas. In the technology area of Autonomy and Intelligent Decision Aiding, Dr. Guffey initiated the first in-house investigations into the use and applicability of Artificial Intelligence Technology at then MCAIR in 1984 and lead the proposal team that won the DARPA Pilot’s Associate Program in the mid 80’s. He was the Technical/Deputy Program Manager of the PA program, which built an AI-based system to assist a fighter pilot in performing future advanced combat missions. The software architecture for PA was the basis for the Rotor Craft Pilot’s Associate system, which in turn was the genus of the Unmanned Combat Air Vehicle and some elements of the Future Combat System program for unmanned vehicles. He was a member of the Integration team for the Unmanned Systems organization when it was formed in 2001 and responsible for the development of a definition of potential future markets that could capitalize on advanced automation and autonomous applications for platforms and systems. In 2004 Jim acted as the Boeing management/technical liaison to the Carnegie Mellon University (CMU) Red. Team for the DARPA Grand Challenge autonomous vehicle race in the deserts of California and Nevada. Part of that role was to be the broker to reach back into Boeing for technical help requested by the onsite Boeing employees at CMU to make the vehicles successful. MSG 06 -081332 -002. ppt

I’ve Been at This a Long Time Autonomy Intelligence Ether Common Sense Recent Introductions by a Colleague: “Jim’s Been at This for Decades!” Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -004. ppt

And……, I Guess It’s True! Absolu tely Insan e ial tific nce Ar lige ntel I Asinin e Ideas mal Ani cts tin Ins Awe g Inspirin Acute ion Indigest Actio n Item Abandon ed Ideas nced Adva nity Insa Almost ible Incred Circa 1985 Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -005. ppt

A Few Definitions Intelligence: • The Ability to Comprehend; to Understand Profit From Experience • The Capacity to Acquire/Apply Knowledge, Especially Toward a Purposeful Goal • The Faculty of Thought and Reason • An Intelligent, Incorporeal Being, Especially an Angel Smart: • Showing Mental Alertness and Calculation and Resourcefulness • Of, Relating to, or Being a Highly Automated Device, Especially One That Imitates Human Intelligence; Does the Right Thing in a Wide Variety of Complicated Circumstances Autonomous: • Not Controlled By Others or By Outside Forces; Independent • Independent in Mind or Judgment; Self-Directed Automation: • Without Volition or Conscious Control; Self-Regulating (i. e. Washing Machine) • Automatic, as Opposed to Human, Operation or Control of a Process, Equipment or a System; or the Techniques and Equipment Used to Achieve This Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -026. ppt

Definition of Ether: “An All-Pervading, Infinitely Elastic, Massless Medium Formerly Postulated as the Medium of Propagation of Electromagnetic Waves; But Subsequently Found to Be Nonexistent. ” Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -006. ppt

The Dawn of Artificial Intelligence • Alan Turing’s Test; 1950 – What He Claimed Was That With Intelligence, Imitation Is as Good as the Real Thing • The Quest for “Artificial Intelligence” Started in ~ 1955 I propose to consider the question, "Can Machines Think? " - Alan Mathison Turing (1950) Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -027. ppt

Hard Artificial Intelligence. The Holy Grail Sonny The Terminator HAL 9000 Knight Rider (Kitt) Iron Giant R 2 D 2 Johnny Five Robby the Robot Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -010. ppt

The Ups and Downs of the Journey Icarus: The First AI Guy • Over Ambitious Expectations → • Disillusion → • AI Winters in 70’s and 90’s • AI Expectations – Maybe Time for Another Up Swing? Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -009. ppt

It Doesn’t Help That the Definition Keeps Changing At Every Step of the Way, on the Quest for Hard AI Systems, Whatever We Make Work, We Immediately Declare That It Is Not Intelligent ? ? No Intelligent ? ? No System A System D System C System B s i creas In ng C itie pabil a The Creator Knows It Isn’t Smart, Even If the Audience Thinks It Is! (So maybe We Should Get Rid of the Creators After They Create? ) Why? Because Once We See What We Did to Make It Work We Realize That It Is Only Doing What We Told It to Do! Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -011. ppt

Why Do Today’s Systems Look “Smarter”? • Brute Force Approach With Elegance – Storage + Throughput • These Allow Much More Complicated Algorithms to Be Imploded, But the Systems Aren’t Any Smarter in Human Perspective Than Before – We Ultimately Are Able to Significantly Expand the Number and Levels of Contingencies That We Can Ask a System to Address + • (e. g. Deep Blue Chess Champion 60, 000, 000 moves in 180 seconds) It’s All About Contingencies! Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -012. ppt

INCOSE Mandatory Equation E= 2 mc Ether = man chasing conundrums* *Conundrum: A paradoxical, insoluble, or difficult problem; a dilemma. Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -014. ppt

A Working Definition of an Intelligent System Three Easy Steps 1. 2. Take Any “Smart” System of Today -barian Locate a Time Machine 3. Go Back in Time and Ask the Locals if They Think It Is Intelligent? If They Say Yes; It’s Because It Appears to Them to Be So Far Ahead of What They Could Do With the Systems of Their Day That it Must Be “Intelligent” Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -015. ppt

Be Careful of What You Ask For ! If We Make a System that Thinks Like a Human and is Truly Artificially Intelligent……. Sonny HAL 9000 Whoops…, Sorry , Dave ! …. Then, Like Humans, it Could Also be Artificially Stupid ** ! Do We Really Want Them to Be Just Like Us ? Copyright © 2005 Boeing. All rights reserved. ** Compliments to Jeff Erickson MSG 06 -081332 -017. ppt

Well, Gee it’s only a small chlorine gas leak It’s Not Like We Always Make Good Decisions! If one lift won’t Why use Who says I make it, when Cleats can’t reach I’ve got this Just use two ! this lightbulb? cool bucket ? Hmmmm Aluminum, Electricity, and Bare Feet. Just Add Water ! And thedon’twondered Why he got Fired? ? guy need no Well, Gee Iit’s only a smallstinking jack !leak chlorine gas Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -038. ppt

Focused on the Wrong Thing I Think I am. Therefore, I Am. I Think? Moody Blues • Debates About Intelligence Are Interesting But They Don’t Get You Much • We in an Engineering Company Look at It From the Practical Perspective • See If We Can Make Some Money Out of It Along the Way! • We Still Have to Stay in Business While We Are Figuring It Out Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -017. ppt

What Our Customers Really Want is… • The Endstate Capability to Handle More and More Complicated, Dynamic Situations for Their Product or Service That Provide a Safe and Acceptable, Real Time Solution With Minimal Human Involvement • Whether You Call It Smart, Robust, Multiple Layers of Contingency Management or Whatever Doesn’t Matter DANTE II – Increasing Capability with Increasing Autonomy Is What Most Customers Ultimately Want Personal Satellite Assistant The Line Between Automation, Autonomy, Smartness, and Intelligence Doesn’t Really Matter! Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -018. ppt

Autonomous Control Level (ACL) Chart Observe Act Decide Orient 10 9 8 PA 7 NG SI EA R 6 INC F 5 Group Accomplishment of Strategic goal with No Supervisory Assistance EO PL AM 4 3 CA IES T ILI B EX 2 1 0 Remotely Piloted Vehicles Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -031. ppt

Autonomy Allocation Problem Decomposition 1. Complexity Defines the Total Problem Solving Requirements (TPSR) 100 Examples: Program (X, Y) Mission Complexity (Y) U 2 (10, 10) (GTAS) B 2 (20, 20) F 22 (50, 50) UUV (10, 10) UCAR (70, 70) Orbital Express (40, 40) PTS (10, 80) etc, . . Problem Solving Requirements Environment Complexity (X) 100 The Combination of Mission and Environment Complexities Determine the Magnitude of the “Total Problem Solving Requirements” Copyright © 2005 Boeing. All rights reserved. MSG 06 -081332 -029. ppt

Autonomy Allocation Problem Decomposition 2. Who Does What, …When? The Smaller the Arrow the Higher the Chance of Automating the Solution and Increasing the Autonomy of the Platform/System Human 100 Total Problem Solving Requirement • Mission and environment complexities determine the total problem solving requirements eg. The Mars Rover Boys Spirit and Opportunity are highly Autonomous 100 Platform/System Network Copyright © 2005 Boeing. All rights reserved. 100 Level of Autonomy to Platform or System Is Both Driven and Constrained by Customer Requirements …and Bounded by Technology MSG 06 -081332 -030. ppt

So, Can We Ever Create a Hard Artificially Intelligent System? • Answer: Alas, ……No Make the World a Little Better Along the Way!! • But Who Cares? ? We Can Still Make Them SMARTER ! It’s Still a Terrific Idea to Keep Striving for Real (? ) Artificial Intelligence If It Looks Like a Duck and Walks Like a …. Copyright © 2005 Boeing. All rights reserved. Vaucanson’s Duck MSG 06 -081332 -019. ppt