Jump to content

Neats and scruffies: Difference between revisions

From Wikipedia, the free encyclopedia
Content deleted Content added
"The victory of neats" in the 1990s: Although Bayesian networks were among the first of the new "mathematical" tools used by AI, they did not prove to be the most important.
Cewbot (talk | contribs)
m Fixing broken anchor: Reminder of an inactive anchor: initiated in 1984
 
(33 intermediate revisions by 13 users not shown)
Line 1: Line 1:
{{short description|Forms of artificial intelligence research}}
{{short description|Forms of artificial intelligence research}}
'''Neat''' and '''scruffy''' are two contrasting approaches to [[artificial intelligence]] (AI) research. The distinction was made in the 70s and was a subject of discussion until the middle 80s. In the 1990s and 21st century AI research adopted "neat" approaches almost exclusively and these have proven to be the most successful.{{sfn|Russell|Norvig|2003|p=25-26}}{{sfn|McCorduck|2004|p=487}}
In the [[history of artificial intelligence]] (AI), '''neat''' and '''scruffy''' are two contrasting approaches to AI research. The distinction was made in the 1970s, and was a subject of discussion until the mid-1980s.{{sfn|McCorduck|2004|pp=421–424, 486–489}}{{sfn|Crevier|1993|p=168}}{{sfn|Nilsson|1983|pp=10–11}}


"Neats" use algorithms based on formal paradigms such as [[logic]], [[mathematical optimization]] or [[neural networks]]. Neat researchers and analysts have expressed the hope that a single formal paradigm can be extended and improved to achieve [[artificial general intelligence|general intelligence]] and [[superintelligence]].
"Neats" use algorithms based on a single formal paradigm, such as [[logic]], [[mathematical optimization]], or [[Neural network (machine learning)|neural networks]]. Neats verify their programs are correct via rigorous mathematical theory. Neat researchers and analysts tend to express the hope that this single formal paradigm can be extended and improved in order to achieve [[Artificial general intelligence|general intelligence]] and [[superintelligence]].


"Scruffies" use any number of different algorithms and methods to achieve intelligent behavior. Scruffy programs may require large amounts of [[hand coding]] or [[knowledge engineering]]. Scruffies have argued that the general intelligence can only implemented by solving a large number of essentially unrelated problems, and that there is no [[wikt:magic bullet|magic bullet]] that will allow programs to develop general intelligence autonomously.
"Scruffies" use any number of different algorithms and methods to achieve intelligent behavior, and rely on incremental testing to verify their programs. Scruffy programming requires large amounts of [[hand coding]] and [[knowledge engineering]]. Scruffy experts have argued that general intelligence can only be implemented by solving a large number of essentially unrelated problems, and that there is no [[silver bullet]] that will allow programs to develop general intelligence autonomously.


The neat approach is similar to physics, in that it uses simple mathematical models as its foundation. The scruffy approach is more like biology, where much of the work involves studying and categorizing diverse phenomena.{{efn|name="chomsky"
[[John Brockman (literary agent)|John Brockman]] compares the neat approach to [[physics]], in that it uses simple mathematical models as its foundation. The scruffy approach is more [[Biology|biological]], in that much of the work involves studying and categorizing diverse phenomena.{{efn|name="chomsky"
|[[John Brockman (literary agent)|John Brockman]] writes "Chomsky has always adopted the physicist's philosophy of science, which is that you have hypotheses you check out, and that you could be wrong. This is absolutely antithetical to the AI philosophy of science, which is much more like the way a biologist looks at the world. The biologist's philosophy of science says that human beings are what they are, you find what you find, you try to understand it, categorize it, name it, and organize it. If you build a model and it doesn't work quite right, you have to fix it. It's much more of a "discovery" view of the world."{{sfn|Brockman|1996|loc=[https://www.edge.org/conversation/information-is-surprises Chapter 9: Information is Surprises]}}
|[[John Brockman (literary agent)|John Brockman]] writes "Chomsky has always adopted the physicist's philosophy of science, which is that you have hypotheses you check out, and that you could be wrong. This is absolutely antithetical to the AI philosophy of science, which is much more like the way a biologist looks at the world. The biologist's philosophy of science says that human beings are what they are, you find what you find, you try to understand it, categorize it, name it, and organize it. If you build a model and it doesn't work quite right, you have to fix it. It's much more of a "discovery" view of the world."{{sfn|Brockman|1996|loc=[https://www.edge.org/conversation/information-is-surprises Chapter 9: Information is Surprises]}}
}}
}}

Modern AI has elements of both scruffy and neat approaches. Scruffy AI researchers in the 1990s applied mathematical rigor to their programs, as neat experts did.{{sfn|Russell|Norvig|2021|p=24}}{{sfn|McCorduck|2004|p=487}} They also express the hope that there is a single paradigm (a "master algorithm") that will cause general intelligence and superintelligence to emerge.{{sfn|Domingos|2015}} But modern AI also resembles the scruffies:{{sfn|Russell|Norvig|2021|p=26}} modern [[machine learning]] applications require a great deal of hand-tuning and incremental testing; while the general algorithm is mathematically rigorous, accomplishing the specific goals of a particular application is not. Also, in the early 2000s, the field of [[software development]] embraced [[extreme programming]], which is a modern version of the scruffy methodology: try things and test them, without wasting time looking for more elegant or general solutions.


==Origin in the 1970s==
==Origin in the 1970s==
The distinction between neat and scruffy originated in the mid-1970s, by [[Roger Schank]]. Schank used the terms to characterize the difference between his work on [[natural language processing]] (which represented [[Commonsense knowledge (artificial intelligence)|commonsense knowledge]] in the form of large amorphous [[semantic networks]]) from the work of [[John McCarthy (computer scientist)|John McCarthy]], [[Allen Newell]], [[Herbert A. Simon]], [[Robert Kowalski]] and others whose work was based on logic and formal extensions of logic.{{sfn|Crevier|1993|p=168}} Schank described himself as an AI scruffy. He made this distinction in linguistics, arguing strongly against Chomsky's view of language.{{efn|name="chomsky"}}
The distinction between neat and scruffy originated in the mid-1970s, by [[Roger Schank]]. Schank used the terms to characterize the difference between his work on [[natural language processing]] (which represented [[Commonsense knowledge (artificial intelligence)|commonsense knowledge]] in the form of large amorphous [[semantic networks]]) from the work of [[John McCarthy (computer scientist)|John McCarthy]], [[Allen Newell]], [[Herbert A. Simon]], [[Robert Kowalski]] and others whose work was based on logic and formal extensions of logic.{{sfn|Crevier|1993|p=168}} Schank described himself as an AI scruffy. He made this distinction in linguistics, arguing strongly against Chomsky's view of language.{{efn|name="chomsky"}}


The distinction was also partly geographical and cultural: "scruffy" attributes were exemplified by AI research at [[MIT]] under [[Marvin Minsky]] in the 1970s. The laboratory was famously "freewheeling" and researchers often developed AI programs by spending long hours fine-tuning programs until they showed the required behavior. Important and influential "scruffy" programs developed at MIT included [[Joseph Weizenbaum]]'s [[ELIZA]], which behaved as if it spoke English, without any formal knowledge at all, and [[Terry Winograd]]'s{{efn|Winograd also became a critic of early approaches to AI as well, arguing that intelligent machines could not be built using formal symbols exclusively, but required [[embodied cognition]].{{sfn|Winograd|Flores|1986}}}} [[SHRDLU]], which could successfully answer queries and carry out actions in a simplified world consisting of blocks and a robot arm.{{sfn|Crevier|1993|pp=84−102}}{{sfn|Russell|Norvig|2003|p=19}} SHRDLU, while successful, could not be scaled up into a useful natural language processing system, because it lacked a structured design. Maintaining a larger version of the program proved to be impossible, i.e. it was too scruffy to be extended.
The distinction was also partly geographical and cultural: "scruffy" attributes were exemplified by AI research at [[MIT]] under [[Marvin Minsky]] in the 1970s. The laboratory was famously "freewheeling" and researchers often developed AI programs by spending long hours fine-tuning programs until they showed the required behavior. Important and influential "scruffy" programs developed at MIT included [[Joseph Weizenbaum]]'s [[ELIZA]], which behaved as if it spoke English, without any formal knowledge at all, and [[Terry Winograd]]'s{{efn|Winograd also became a critic of early approaches to AI as well, arguing that intelligent machines could not be built using formal symbols exclusively, but required [[embodied cognition]].{{sfn|Winograd|Flores|1986}}}} [[SHRDLU]], which could successfully answer queries and carry out actions in a simplified world consisting of blocks and a robot arm.{{sfn|Crevier|1993|pp=84−102}}{{sfn|Russell|Norvig|2021|p=20}} SHRDLU, while successful, could not be scaled up into a useful natural language processing system, because it lacked a structured design. Maintaining a larger version of the program proved to be impossible, i.e. it was too scruffy to be extended.


Other AI laboratories (of which the largest were [[Stanford]], [[Carnegie Mellon University]] and the [[University of Edinburgh]]) focused on logic and formal problem solving as a basis for AI. These institutions supported the work of John McCarthy, Herbert Simon, Allen Newell, [[Donald Michie]], Robert Kowalski, and other "neats".
Other AI laboratories (of which the largest were [[Stanford]], [[Carnegie Mellon University]] and the [[University of Edinburgh]]) focused on logic and formal problem solving as a basis for AI. These institutions supported the work of John McCarthy, Herbert Simon, Allen Newell, [[Donald Michie]], Robert Kowalski, and other "neats".


The contrast between [[MIT]]'s approach and other laboratories was also described as a "procedural/declarative distinction". Programs like SHRDLU were designed as agents that carried out actions. They executed "procedures". Other programs were designed as inference engines that manipulated formal statements (or "declarations") about the world and translated these manipulations into actions. In his 1983 presidential address to [[Association for the Advancement of Artificial Intelligence]], [[Nils Nilsson (researcher)|Nils Nilsson]] discussed the issue, arguing that "the field needed both". He wrote "much of the knowledge we want our programs to have can and should be represented declaratively in some kind of declarative, logic-like formalism. Ad hoc structures have their place, but most of these come from the domain itself." Alex P. Pentland and Martin Fischler of [[MIT]] concurred about the anticipated role of deduction and logic-like formalisms in future AI research, but not to the extent that Nilsson described.<ref>Pentland and Fischler 1983, quoted in {{Harvnb|McCorduck|2004|pp=421–424}}</ref>
The contrast between [[MIT]]'s approach and other laboratories was also described as a "procedural/declarative distinction". Programs like SHRDLU were designed as agents that carried out actions. They executed "procedures". Other programs were designed as inference engines that manipulated formal statements (or "declarations") about the world and translated these manipulations into actions.
In his 1983 presidential address to [[Association for the Advancement of Artificial Intelligence]], [[Nils Nilsson (researcher)|Nils Nilsson]] discussed the issue, arguing that "the field needed both". He wrote "much of the knowledge we want our programs to have can and should be represented declaratively in some kind of declarative, logic-like formalism. Ad hoc structures have their place, but most of these come from the domain itself." Alex P. Pentland and Martin Fischler of [[SRI International]] concurred about the anticipated role of deduction and logic-like formalisms in future AI research, but not to the extent that Nilsson described.<ref>Pentland and Fischler 1983, quoted in {{Harvnb|McCorduck|2004|pp=421–424}}</ref>


==Scruffy projects in the 1980s==
==Scruffy projects in the 1980s==
Line 23: Line 27:
The scruffy approach was applied to robotics by [[Rodney Brooks]] in the mid-1980s. He advocated building robots that were, as he put it, [[Fast, Cheap and Out of Control]], the title of a 1989 paper co-authored with Anita Flynn. Unlike earlier robots such as [[Shakey the robot|Shakey]] or the Stanford cart, they did not build up representations of the world by analyzing visual information with algorithms drawn from mathematical [[machine learning]] techniques, and they did not plan their actions using formalizations based on logic, such as the '[[Planner programming language|Planner]]' language. They simply reacted to their sensors in a way that tended to help them survive and move.{{sfn|McCorduck|2004|pp=454–459}}
The scruffy approach was applied to robotics by [[Rodney Brooks]] in the mid-1980s. He advocated building robots that were, as he put it, [[Fast, Cheap and Out of Control]], the title of a 1989 paper co-authored with Anita Flynn. Unlike earlier robots such as [[Shakey the robot|Shakey]] or the Stanford cart, they did not build up representations of the world by analyzing visual information with algorithms drawn from mathematical [[machine learning]] techniques, and they did not plan their actions using formalizations based on logic, such as the '[[Planner programming language|Planner]]' language. They simply reacted to their sensors in a way that tended to help them survive and move.{{sfn|McCorduck|2004|pp=454–459}}


[[Douglas Lenat]]'s [[Cyc]] project was [[Cyc#Overview|initiated in 1984]] one of earliest and most ambitious projects to capture all of human knowledge in machine readable form, is "a determinedly scruffy enterprise".{{sfn|McCorduck|2004|p=489}} The Cyc database contains millions of facts about all the complexities of the world, each of which must be entered one at a time, by knowledge engineers. Each of these entries is an ad hoc addition to the intelligence of the system. While there may be a "neat" solution to the problem of commonsense knowledge (such as machine learning algorithms with natural language processing that could study the text available over the internet), no such project has yet been successful.
[[Douglas Lenat]]'s [[Cyc]] project was [[Cyc#Overview|initiated in 1984]]{{Broken anchor|date=2024-12-16|bot=User:Cewbot/log/20201008/configuration|target_link=Cyc#Overview|reason= The anchor (Overview) [[Special:Diff/1241345677|has been deleted]].}} one of earliest and most ambitious projects to capture all of human knowledge in machine readable form, is "a determinedly scruffy enterprise".{{sfn|McCorduck|2004|p=489}} The Cyc database contains millions of facts about all the complexities of the world, each of which must be entered one at a time, by knowledge engineers. Each of these entries is an ad hoc addition to the intelligence of the system. While there may be a "neat" solution to the problem of commonsense knowledge (such as machine learning algorithms with natural language processing that could study the text available over the internet), no such project has yet been successful.


== The Society of Mind ==
== The Society of Mind ==
{{Main|Society of Mind}}
{{Main|Society of Mind}}


In 1986 [[Marvin Minsky]] published ''[[The Society of Mind]]'' which advocated a view of [[intelligence]] and the [[mind]] as an interacting community of [[modularity|modules]] or [[intelligent agent|agents]] that each handled different aspects of cognition, where some modules were specialized for very specific tasks (e.g. [[edge detection]] in the visual cortex) and other modules were specialized to manage communication and prioritization (e.g. [[planning]] and [[attention]] in the frontal lobes). Minsky presented this paradigm as a model of both biological human intelligence and as a blueprint for future work in AI.
In 1986 [[Marvin Minsky]] published ''The Society of Mind'' which advocated a view of [[intelligence]] and the [[mind]] as an interacting community of [[modularity|modules]] or [[intelligent agent|agents]] that each handled different aspects of cognition, where some modules were specialized for very specific tasks (e.g. [[edge detection]] in the visual cortex) and other modules were specialized to manage communication and prioritization (e.g. [[planning]] and [[attention]] in the frontal lobes). Minsky presented this paradigm as a model of both biological human intelligence and as a blueprint for future work in AI.


This paradigm is explicitly "scruffy" in that it does not expect there to be a single algorithm that can be applied to all of the tasks involved in intelligent behavior. He wrote:
This paradigm is explicitly "scruffy" in that it does not expect there to be a single algorithm that can be applied to all of the tasks involved in intelligent behavior.{{sfn|Crevier|1993|p=254}} Minsky wrote:
{{quote|What magical trick makes us intelligent? The trick is that there is no trick. The power of intelligence stems from our vast diversity, not from any single, perfect principle.{{sfn|Minsky|1986|p=308}}}}
{{quote|What magical trick makes us intelligent? The trick is that there is no trick. The power of intelligence stems from our vast diversity, not from any single, perfect principle.{{sfn|Minsky|1986|p=308}}}}
As of 1991, Minsky was still publishing papers evaluating the relative advantages of the neat versus scruffy approaches, e.g. “Logical Versus Analogical or Symbolic Versus Connectionist or Neat Versus
As of 1991, Minsky was still publishing papers evaluating the relative advantages of the neat versus scruffy approaches, e.g. “Logical Versus Analogical or Symbolic Versus Connectionist or Neat Versus
Scruffy”.{{sfn|Lehnert|1994}}
Scruffy”.{{sfn|Lehnert|1994}}


== "The victory of neats" in the 1990s ==
== Modern AI as both neat and scruffy ==
New [[Artificial intelligence#Statistical|statistical]] and mathematical approaches to AI were developed in the 1990s, using highly developed formalisms such as [[optimization (mathematics)|mathematical optimization]] and [[neural networks]]. This general trend towards more formal methods in AI is described as "the victory of the neats" by [[Peter Norvig]] and [[Stuart J. Russell|Stuart Russell]].{{sfn|Russell|Norvig|2003|p=25−26}} [[Pamela McCorduck]] wrote that "As I write, AI enjoys a Neat hegemony, people who believe that machine intelligence, at least, is best expressed in logical, even mathematical terms."{{sfn|McCorduck|2004|p=487}}
New [[Artificial intelligence#Statistical|statistical]] and mathematical approaches to AI were developed in the 1990s, using highly developed formalisms such as [[optimization (mathematics)|mathematical optimization]] and [[Artificial neural network|neural networks]]. [[Pamela McCorduck]] wrote that "As I write, AI enjoys a Neat hegemony, people who believe that machine intelligence, at least, is best expressed in logical, even mathematical terms."{{sfn|McCorduck|2004|p=487}} This general trend towards more formal methods in AI was described as "the victory of the neats" by [[Peter Norvig]] and [[Stuart J. Russell|Stuart Russell]] in 2003.{{sfn|Russell|Norvig|2003|p=25−26}}


However, by 2021, Russell and Norvig had changed their minds.{{sfn|Russell|Norvig|2021|p=23}} Deep learning networks and machine learning in general require extensive fine tuning -- they must be iteratively tested until they begin to show the desired behavior. This is a scruffy methodology.
Neat solutions to problems such as machine learning and [[computer vision]] have been highly successful in the 21st century.{{sfn|Russell|Norvig|2003|p=25−26}} These solutions, however, have mostly been applied to specific problems with specific solutions, and the problem of [[Artificial general intelligence]] (AGI) remains unsolved.

The terms "neat" and "scruffy" are not often used by AI researchers in the 21st century, with some exceptions such as talks by [[Karl Friston]] on the [[Free energy principle]], where he refers to physicists as "Neats" and AI researchers as "Scruffies" (and philosophers as "Mystics").


==Well-known examples==
==Well-known examples==
Line 64: Line 66:


* [[History of artificial intelligence]]
* [[History of artificial intelligence]]
* {{section link|Artificial Intelligence|Historical approaches}}
* [[Soft computing]]
* [[Soft computing]]
* [[Symbolic AI]]
* [[Symbolic AI]]
Line 77: Line 78:
* {{cite book |last1=Brockman |first1=John |title=Third Culture: Beyond the Scientific Revolution |date=7 May 1996 |publisher=Simon and Schuster |url=https://www.amazon.com/exec/obidos/ASIN/0684823446/qid=913732847/sr=1-3/002-6796862-3062667 |access-date=2 August 2021 |language=en}}
* {{cite book |last1=Brockman |first1=John |title=Third Culture: Beyond the Scientific Revolution |date=7 May 1996 |publisher=Simon and Schuster |url=https://www.amazon.com/exec/obidos/ASIN/0684823446/qid=913732847/sr=1-3/002-6796862-3062667 |access-date=2 August 2021 |language=en}}
* {{Crevier 1993}}.
* {{Crevier 1993}}.
* {{cite book
| last = Domingos | first = Pedro | author-link = Pedro Domingos
| title = The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World
| date = 22 September 2015
| publisher = [[Basic Books]]
| isbn = 978-0465065707
}}
* {{cite book
* {{cite book
|last=Lehnert|first=Wendy C.
|last=Lehnert|first=Wendy C.
Line 84: Line 92:
|publisher=Taylor & Francis Group |location=New York, NY |page=150 |edition=First |url=https://www.taylorfrancis.com/books/edit/10.4324/9780203773574/beliefs-reasoning-decision-making-roger-schank-ellen-langer |access-date=2 August 2021
|publisher=Taylor & Francis Group |location=New York, NY |page=150 |edition=First |url=https://www.taylorfrancis.com/books/edit/10.4324/9780203773574/beliefs-reasoning-decision-making-roger-schank-ellen-langer |access-date=2 August 2021
|language=en
|language=en
|chapter=5: Cognition, Computers, and Car Bombs: How Yale Prepared Me for the 90’s}}
|chapter=5: Cognition, Computers, and Car Bombs: How Yale Prepared Me for the 90’s|doi=10.4324/9780203773574
|isbn=9781134781621
}}
* {{cite book
* {{cite book
|last1=Minsky|first1=Marvin
|last1=Minsky|first1=Marvin
Line 92: Line 102:
|url=https://archive.org/details/societyofmind00marv}}
|url=https://archive.org/details/societyofmind00marv}}
* {{McCorduck 2004}}.
* {{McCorduck 2004}}.
<!-- Note: this article needs both new and old editions of R&N, because it discusses how they changed their minds -->
* {{Russell Norvig 2003}}.
* {{Cite book
| first1 = Stuart J. | last1 = Russell | author1-link=Stuart J. Russell
| first2 = Peter | last2 = Norvig | author2-link=Peter Norvig
| title = Artificial Intelligence: A Modern Approach
| year = 2003
| edition = 2nd
| publisher = Prentice Hall | publication-place = Upper Saddle River, New Jersey
| isbn = 0-13-790395-2
}}
* {{Cite book
| first1 = Stuart J. | last1 = Russell | author1-link=Stuart J. Russell
| first2 = Peter | last2 = Norvig | author2-link=Peter Norvig
| title =[[Artificial Intelligence: A Modern Approach]]
| year = 2021
| edition = 4th
| isbn = 9780134610993
| lccn = 20190474
| publisher = Pearson | location = Hoboken
}}
* {{cite book
* {{cite book
| last1 = Winograd | first1 = Terry
| last1 = Winograd | first1 = Terry
Line 126: Line 155:
| pmid = 11201756
| pmid = 11201756
| issue = 6818
| issue = 6818
| bibcode = 2001Natur.409..409B
| s2cid = 4430614
| s2cid = 4430614
| doi-access = free
| doi-access = free
}}
}}
*{{FOLDOC}}


{{DEFAULTSORT:Neats Vs. Scruffies}}
{{DEFAULTSORT:Neats Vs. Scruffies}}
[[Category:Philosophy of artificial intelligence]]
[[Category:Philosophy of artificial intelligence]]
[[Category:Artificial intelligence]]

Latest revision as of 00:00, 16 December 2024

In the history of artificial intelligence (AI), neat and scruffy are two contrasting approaches to AI research. The distinction was made in the 1970s, and was a subject of discussion until the mid-1980s.[1][2][3]

"Neats" use algorithms based on a single formal paradigm, such as logic, mathematical optimization, or neural networks. Neats verify their programs are correct via rigorous mathematical theory. Neat researchers and analysts tend to express the hope that this single formal paradigm can be extended and improved in order to achieve general intelligence and superintelligence.

"Scruffies" use any number of different algorithms and methods to achieve intelligent behavior, and rely on incremental testing to verify their programs. Scruffy programming requires large amounts of hand coding and knowledge engineering. Scruffy experts have argued that general intelligence can only be implemented by solving a large number of essentially unrelated problems, and that there is no silver bullet that will allow programs to develop general intelligence autonomously.

John Brockman compares the neat approach to physics, in that it uses simple mathematical models as its foundation. The scruffy approach is more biological, in that much of the work involves studying and categorizing diverse phenomena.[a]

Modern AI has elements of both scruffy and neat approaches. Scruffy AI researchers in the 1990s applied mathematical rigor to their programs, as neat experts did.[5][6] They also express the hope that there is a single paradigm (a "master algorithm") that will cause general intelligence and superintelligence to emerge.[7] But modern AI also resembles the scruffies:[8] modern machine learning applications require a great deal of hand-tuning and incremental testing; while the general algorithm is mathematically rigorous, accomplishing the specific goals of a particular application is not. Also, in the early 2000s, the field of software development embraced extreme programming, which is a modern version of the scruffy methodology: try things and test them, without wasting time looking for more elegant or general solutions.

Origin in the 1970s

[edit]

The distinction between neat and scruffy originated in the mid-1970s, by Roger Schank. Schank used the terms to characterize the difference between his work on natural language processing (which represented commonsense knowledge in the form of large amorphous semantic networks) from the work of John McCarthy, Allen Newell, Herbert A. Simon, Robert Kowalski and others whose work was based on logic and formal extensions of logic.[2] Schank described himself as an AI scruffy. He made this distinction in linguistics, arguing strongly against Chomsky's view of language.[a]

The distinction was also partly geographical and cultural: "scruffy" attributes were exemplified by AI research at MIT under Marvin Minsky in the 1970s. The laboratory was famously "freewheeling" and researchers often developed AI programs by spending long hours fine-tuning programs until they showed the required behavior. Important and influential "scruffy" programs developed at MIT included Joseph Weizenbaum's ELIZA, which behaved as if it spoke English, without any formal knowledge at all, and Terry Winograd's[b] SHRDLU, which could successfully answer queries and carry out actions in a simplified world consisting of blocks and a robot arm.[10][11] SHRDLU, while successful, could not be scaled up into a useful natural language processing system, because it lacked a structured design. Maintaining a larger version of the program proved to be impossible, i.e. it was too scruffy to be extended.

Other AI laboratories (of which the largest were Stanford, Carnegie Mellon University and the University of Edinburgh) focused on logic and formal problem solving as a basis for AI. These institutions supported the work of John McCarthy, Herbert Simon, Allen Newell, Donald Michie, Robert Kowalski, and other "neats".

The contrast between MIT's approach and other laboratories was also described as a "procedural/declarative distinction". Programs like SHRDLU were designed as agents that carried out actions. They executed "procedures". Other programs were designed as inference engines that manipulated formal statements (or "declarations") about the world and translated these manipulations into actions.

In his 1983 presidential address to Association for the Advancement of Artificial Intelligence, Nils Nilsson discussed the issue, arguing that "the field needed both". He wrote "much of the knowledge we want our programs to have can and should be represented declaratively in some kind of declarative, logic-like formalism. Ad hoc structures have their place, but most of these come from the domain itself." Alex P. Pentland and Martin Fischler of SRI International concurred about the anticipated role of deduction and logic-like formalisms in future AI research, but not to the extent that Nilsson described.[12]

Scruffy projects in the 1980s

[edit]

The scruffy approach was applied to robotics by Rodney Brooks in the mid-1980s. He advocated building robots that were, as he put it, Fast, Cheap and Out of Control, the title of a 1989 paper co-authored with Anita Flynn. Unlike earlier robots such as Shakey or the Stanford cart, they did not build up representations of the world by analyzing visual information with algorithms drawn from mathematical machine learning techniques, and they did not plan their actions using formalizations based on logic, such as the 'Planner' language. They simply reacted to their sensors in a way that tended to help them survive and move.[13]

Douglas Lenat's Cyc project was initiated in 1984[broken anchor] one of earliest and most ambitious projects to capture all of human knowledge in machine readable form, is "a determinedly scruffy enterprise".[14] The Cyc database contains millions of facts about all the complexities of the world, each of which must be entered one at a time, by knowledge engineers. Each of these entries is an ad hoc addition to the intelligence of the system. While there may be a "neat" solution to the problem of commonsense knowledge (such as machine learning algorithms with natural language processing that could study the text available over the internet), no such project has yet been successful.

The Society of Mind

[edit]

In 1986 Marvin Minsky published The Society of Mind which advocated a view of intelligence and the mind as an interacting community of modules or agents that each handled different aspects of cognition, where some modules were specialized for very specific tasks (e.g. edge detection in the visual cortex) and other modules were specialized to manage communication and prioritization (e.g. planning and attention in the frontal lobes). Minsky presented this paradigm as a model of both biological human intelligence and as a blueprint for future work in AI.

This paradigm is explicitly "scruffy" in that it does not expect there to be a single algorithm that can be applied to all of the tasks involved in intelligent behavior.[15] Minsky wrote:

What magical trick makes us intelligent? The trick is that there is no trick. The power of intelligence stems from our vast diversity, not from any single, perfect principle.[16]

As of 1991, Minsky was still publishing papers evaluating the relative advantages of the neat versus scruffy approaches, e.g. “Logical Versus Analogical or Symbolic Versus Connectionist or Neat Versus Scruffy”.[17]

Modern AI as both neat and scruffy

[edit]

New statistical and mathematical approaches to AI were developed in the 1990s, using highly developed formalisms such as mathematical optimization and neural networks. Pamela McCorduck wrote that "As I write, AI enjoys a Neat hegemony, people who believe that machine intelligence, at least, is best expressed in logical, even mathematical terms."[6] This general trend towards more formal methods in AI was described as "the victory of the neats" by Peter Norvig and Stuart Russell in 2003.[18]

However, by 2021, Russell and Norvig had changed their minds.[19] Deep learning networks and machine learning in general require extensive fine tuning -- they must be iteratively tested until they begin to show the desired behavior. This is a scruffy methodology.

Well-known examples

[edit]

Neats

Scruffies

See also

[edit]

Notes

[edit]
  1. ^ a b John Brockman writes "Chomsky has always adopted the physicist's philosophy of science, which is that you have hypotheses you check out, and that you could be wrong. This is absolutely antithetical to the AI philosophy of science, which is much more like the way a biologist looks at the world. The biologist's philosophy of science says that human beings are what they are, you find what you find, you try to understand it, categorize it, name it, and organize it. If you build a model and it doesn't work quite right, you have to fix it. It's much more of a "discovery" view of the world."[4]
  2. ^ Winograd also became a critic of early approaches to AI as well, arguing that intelligent machines could not be built using formal symbols exclusively, but required embodied cognition.[9]

Citations

[edit]
  1. ^ McCorduck 2004, pp. 421–424, 486–489.
  2. ^ a b Crevier 1993, p. 168.
  3. ^ Nilsson 1983, pp. 10–11.
  4. ^ Brockman 1996, Chapter 9: Information is Surprises.
  5. ^ Russell & Norvig 2021, p. 24.
  6. ^ a b McCorduck 2004, p. 487.
  7. ^ Domingos 2015.
  8. ^ Russell & Norvig 2021, p. 26.
  9. ^ Winograd & Flores 1986.
  10. ^ Crevier 1993, pp. 84−102.
  11. ^ Russell & Norvig 2021, p. 20.
  12. ^ Pentland and Fischler 1983, quoted in McCorduck 2004, pp. 421–424
  13. ^ McCorduck 2004, pp. 454–459.
  14. ^ McCorduck 2004, p. 489.
  15. ^ Crevier 1993, p. 254.
  16. ^ Minsky 1986, p. 308.
  17. ^ Lehnert 1994.
  18. ^ Russell & Norvig 2003, p. 25−26.
  19. ^ Russell & Norvig 2021, p. 23.

References

[edit]
  • Brockman, John (7 May 1996). Third Culture: Beyond the Scientific Revolution. Simon and Schuster. Retrieved 2 August 2021.
  • Crevier, Daniel (1993). AI: The Tumultuous Search for Artificial Intelligence. New York, NY: BasicBooks. ISBN 0-465-02997-3..
  • Domingos, Pedro (22 September 2015). The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World. Basic Books. ISBN 978-0465065707.
  • Lehnert, Wendy C. (1 May 1994). "5: Cognition, Computers, and Car Bombs: How Yale Prepared Me for the 90's". In Schank, Robert; Langer, Ellen (eds.). Beliefs, Reasoning, and Decision Making: Psycho-Logic in Honor of Bob Abelson (First ed.). New York, NY: Taylor & Francis Group. p. 150. doi:10.4324/9780203773574. ISBN 9781134781621. Retrieved 2 August 2021.
  • Minsky, Marvin (1986). The Society of Mind. New York: Simon & Schuster. ISBN 0-671-60740-5.
  • McCorduck, Pamela (2004), Machines Who Think (2nd ed.), Natick, Massachusetts: A. K. Peters, ISBN 1-5688-1205-1.
  • Russell, Stuart J.; Norvig, Peter (2003). Artificial Intelligence: A Modern Approach (2nd ed.). Upper Saddle River, New Jersey: Prentice Hall. ISBN 0-13-790395-2.
  • Russell, Stuart J.; Norvig, Peter (2021). Artificial Intelligence: A Modern Approach (4th ed.). Hoboken: Pearson. ISBN 9780134610993. LCCN 20190474.
  • Winograd, Terry; Flores (1986). Understanding Computers and Cognition: A New Foundation for Design. Ablex Publ Corp.

Further reading

[edit]