Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered...

16
ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 1 Artificial Intelligence (A.I.):Nosce te ipsum TEC-EDD By Nick Panagiotopoulos Jacobus Neeffes (Dutch, 1610–1660)

Transcript of Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered...

Page 1: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 1

Artificial Intelligence (A.I.):Nosce te ipsum

TEC-EDD

By Nick Panagiotopoulos

Jacobus Neeffes(Dutch, 1610–1660)

Page 2: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 2

What is Artificial Intelligence (A.I.)?

Can we create A.I.?

Methods used by A.I. scientists?

Where A.I. can be used?

How will A.I. affect us?

Page 3: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 3

What is AI?

Artificial intelligence focuses on making already “intelligent” systems capable of simulating human-like decision-making and execution – enabling those systems to perform functions traditionally executed by skilled human professionals – but do so at a much higher level, because of the speed and power available on modern computing platforms.

Artificial intelligence is a branch of computer science that aims to create intelligent machines. (Technopedia)

Artificial intelligence (AI) is an area of computer science that emphasizes the creation of intelligent machines that work and react like humans. (Technopedia)

Artificial intelligence (AI) is the ability of a computer program or a machine to think and learn. (Wikipedia)

The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. (Lexico)

Artificial intelligence (AI), the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. (Encyclopaedia Britannica)

Page 4: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 4

AI is a branch of computer science that is concerned with the automation of intelligent behavior.

What is AI?

Problem with Definition(s).

The problem with “Artificial Intelligence” in general, is the lack of precise, rigor, quantifiable and sound definition of the word “Intelligence”. And since, scientists and engineers are focused on building AI, then having no proper definition is an engineering obstacle due to the fact that no engineering requirements and/or specifications can be generated. No requirements means no design and no tests, no verification & validation plans and procedures (consequently no proof of achieving AI), hence no product or result.

Page 5: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 5

So, can we create AI?

NOT YET.

But, can we create AI in general? Can machines think?

• Computers can only do as they are told and consequently cannot perform original (hence intelligent) actions. Ada Lovelace

• However complicated a machine we construct, it will, if it is a machine, correspond to a formal system, which in turn will be liable to the Gödel procedure for finding a formula unprovable-in-that- system. This formula the machine will be unable to produce as true, although a mind can see that it is true. And so the machine will not be an adequate model of the mind. J.R. Lucas

Page 6: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 6

Quest for the Holy Grail: Implementing A.I.

What can we do?

1) Wait, until we further evolve.

2) Mimic human Intelligent behavior. Sir Galahad - the Quest of the Holy GrailArthur Hughes•Date: 1870•Style: Romanticism

Page 7: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 7

Knowledge Representation & Search

Intelligent activity can be achieved through the use of :

1. Symbol patterns to represent significant aspects of a problem domain.

2. Operations on these patterns to generate potential solutions to problems.

3. Search to select a solution from among these possibilities.

Medium is not important. Even Tinkertoys can achieve intelligent activity.

Page 8: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 8

Representation & Search cont.• Predicate Calculus. It provides the means of describing objects and relations in a problem domain. Through inference rules (i.e. modus ponens) and unification, new knowledge is generated. Then a space full of potential solutions is created and search for the optimum one is performed. Example: A logic based Financial Advisor.

• State Space. Use of graph theory (nodes and arcs) for analyzing the structure and complexity of a problem-solving strategies (i.e. data driven or goal driven strategy). It is a search method (Depth-first and Breadth-first). Example: Travelling Salesperson.

• Heuristic Search. It is an informed guess. Heuristics are fallible. But when there are ambiguities or lack of available data, or exact solution is impossible due to computational cost then heuristic is the best option. Nevertheless, heuristic can lead to a suboptimal solution or fail to find any solution at all. Best-first strategy. It is used in games.

• Expert systems. With the help of human domain experts, the system emulates methodology and performance. But cannot learn. Knowledge must be extracted by humans and encoded in formal language. Not to be confused with cognitive modelling.

Page 9: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 9

Representation & Search cont.

• Reasoning with uncertain or incomplete information. It is common to draw useful conclusions from incomplete and imprecise data with unsound reasoning. For example, medical diagnosis, car problems. In these cases, use of probability theory is appropriate. Bayesian reasoning, Dempster-Shafer Theory, and casual networks are few from the plenty. Fuzzy logic or fuzzy set theory is also part in this domain.

A.I. programming languages: LISP and PROLOG. Python is in the game too.

Utilizing the above methods does not guarantee that the system will possess artificial intelligence. They are just the best we can do now with very good results in many cases.

Page 10: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 10

Where AI can be used?

Overview of AI applications:• Game Playing (i.e. Warfare)• Automated reasoning and theorem proving.• Expert systems• Natural Language understanding and Semantic Modelling.• Modelling Human Performance.• Planning (i.e. Robotics)• Machine Learning.• Parallel Distributed Processing (PDP) and Emergent

Computation.

Page 11: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 11

Machine Learning.

In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

• ML is a subfield of AI.

• It can be manifested through symbol based system. It includes inductive learning, concept learning, inductive bias, similarity-based, explanation-based learning, supervised learning, and unsupervised learning.

• It can also be manifested through connectionist systems or parallel distributed processing (PDP). These are biology or neural inspired approaches. No symbol based. Intelligence arises from complex systems made by simple components (non-linear though) through learning or adaptation. Artificial Neural Networks are included in this category. Important note: there is an overlap between AI, cognitive science, and complex theory.

Page 12: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 12

Machine Learning. cont

• Social and Emergent. Again, intelligence is considered as an emergent property. Evolution is the main driver in this approach (“the survival of the fittest”). It has been formalized by cellular automata, genetic algorithms, genetic programming, artificial life and other forms of emergent computation.

Page 13: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 13

A.I. era! Man-Machine co-existence.Some remarks:

• Artificial Intelligence is another one tool that humans need. A (brain) tool that will assist mankind in tedious mental work, in time consuming mental work, and complex mental works.

• A.I. is not about improving skills or assisting humans in skillful works. Control theory (in general) and signal processing with the advancement of technology can do that.

• A.I. is not about calculations. Computers can do that.

Page 14: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 14

• Machines with A.I. will remain machines that follow orders. So, no apocalyptic scenarios due to A.I.

• Although some relevant jobs will be filled with A.I. machines, more jobs will be created for supporting, building, improving and expanding A.I. technology. There is going to be a market job shift; as it happens whenever a new technology is introduced to the public.

• A.I. is more related with higher level of abstraction and complex tasks.

• A.I. has nothing to do with wisdom, personality, cleverness, character, and self awareness. Therefore, A.I. has no vanity, no preferences, no likes and dislikes, and no (free) will. Artificial Intelligence is just a tool. The way this tool is used depends on the morality of the human user. Like the hammer; someone can build a house or smash a head. Therefore, decision making is controllable.

A.I. era! Man-Machine co-existence.

Page 15: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 15

A.I. era! Man-Machine co-existence.Nevertheless, there are issues that have to considered:

• There is no much problem with the man-machine relation. But there are lots of problems with man-machine-man relation. The necessity of building and including A.I. ethics is growing, and becomes apparent that for the integration of the new technology into humans’ society it will become mandatory.

https://www.globalgovernmentforum.com/french-and-canadians-to-launch-global-panel-on-ethical-ai/

Page 16: Artificial Intelligence (A.I.):Nosce te ipsum€¦ · Machine Learning. In general it is considered that the ability to learn must be a part of any system that claim to exhibit intelligence.

ESA UNCLASSIFIED - For Official Use ESA | 01/01/2016 | Slide 16

THANKS FOR YOUR TIME AND ATTENTION