Ellery Aziel, James Newton Bass, Wolfgang Bibel, David R. Boxall, Gerd Brewka, Carla Brodley, Wilhelm Burger, Joao Cachopo, Norman Carver, Emmanuel Castro, Dan Chisarick, James Coleman, Julie Ann Comparini, Ernest Davis, Tom Dietterich, Chuck Dyer, Doug Edwards, Douglas Fisher, Marek Galecki, Stefan Gerberding, Stuart Gill, Gosta Grahne, Larry Hall, Christoph Herrmann, Magnus Johansson, Gernot Kerschbaumer, Richard Kirby, Gus Larsson, Frank Lee, Edward Lim, Pierre Louveaux, Don Loveland, Sung Myaeng, David McGrane, Jay Mendelsohn, Bernhard Nebel, David Palmer, David Poole, Bill Pringle, William Rapaport, Richard Scherl, Lars Schuster, Soheil Shams, Stuart Shapiro, Jude Shavlik, Bryan So, Andreas Stolcke, Paul Stradling, Jeff Van Baalen, Sunil Vemuri, Michael Wellman, Michael Dean White, Alden Wright,

As of July 1995, AIMA has gone to a second printing (although some books from the first printing may remain on bookstore shelves). The second printing makes corrections on 68 pages. Note that a second printing is quite different from a second edition, which would contain all-new material, not just corrections. (Look for a second edition by the end of 2001 or thereabouts, and see the list of clarifications for some things that will be addressed in the second edition but were too costly to fix in a reprinting.)

If you have a second printing book, you can stop after looking at the Errata for Second Printing. If you have a first printing book, then all the errata below apply to you. You can also see page images of all the corrections.

Chapter 10

• Page 303, Figure 10.3, line -1: Change "{x/var}" to "{var/x}" (This is how it is in the code file logic/unify.lisp)

Chapter 12

• Page 377, line -6: Change "576" to "252". (I don't know where the 576 came from; the computation for 252, (and for 3 × 10³⁰) is shown below:

  CL-USER 1 > (defun f (b s d)
               (loop for i from 1 to d sum (* b (expt s i))))
  F
  
  CL-USER 2 > (f 3 4 3)
  252
  
  CL-USER 3 > (expt 3.0 64)
  3.433684E30
  

Chapter 14

• Page 439, Figure 15.2: Change ".94" to ".95" to match the table on page 438.

Chapter 15

• Page 452, Figure 15.8: The pseudocode for Evidence-Except is incorrect since it omits the case where there is evidence in the Y and Z variables. A corrected postscript version, including additional clarifications, has been added to the list of page images.

Chapter 18

• Page 537, Figure 18.8: The No and Yes branch labels below the Hungry? node should be reversed.

Chapter 22

• Page 668, Second DCG/First-Order Logic table: Change "(s = [sem])" to "(s = [sem]) ^ 0 < sem < 9" (those should be less-than-or-equal signs). Also change "Number(sem,s₁) ^ Digit(sem,s₂)" to "Number(sem₁,s₁) ^ Digit(sem₂,s₂)" (the 1s and 2s are subscripts).

Bibliography

• Page 897: Stolcke (1993) is the wrong reference (it talks about parsing probabilistic context free grammars, whereas we meant to refer to his work on learning such grammars:

  Stolcke, A. (1994). Bayesian Learning of Probabilistic Language Models, PhD thesis, UC Berkeley Computer Science Dept.

Index

• Page 905: Chi-Square is indexed under symbols, but Chi-square pruning is under C. Put them both under symbols.

• Page 912: Add E3 (English subset), 705 708

• Page 919: Indexing of Logic Theorist is inconsistent:

  Logic Theorist, 292
  Logic Theorist (LT), 17
  
  LT, 17
  LT (Logic Theorist), 17
  

  It should be:

  Logic Theorist (LT), 17, 292
  LT (Logic Theorist), 17, 292
  

Binding

Chapter 3

• Page 63, line 14: Change "move the 3" to "move the 4". This reflects a move that is actually possible in the initial state of the diagram.

Chapter 4

• Page 95: line 14: change "32 miles" to "32 kilometers".

• Page 101, Section 4.2, line 6: Change "initial configuration" to "configuration".

• Page 102, lines 3-11: The text does not accurately represent the configurations in Figure 4.7. Change "none of the 8 tiles is in the goal position, so the start state would have h1 = 8" to "seven of the eight tiles are out of position, so the start state would have h1 = 7." Also change "The 8 tiles in the start state give a Manhattan distance of ... = 15" to "Tiles 1 to 8 in the start state give a Manhattan distance of

  h2 = 2 + 3 + 3 + 2 + 4 + 2 + 0 + 2 = 18"

• Page 107, Figure 4.10: The variables f-limit, root and next-f do not need to be static variables. (If they are static the algorithm is still correct, but it holds on to storage needlessly. This problem also occurs on pages 110, 112 and 113.) Change static to local variables.

• Page 110, Figure 4.12, line 3: Change static to local variables. Also, line 4: Change "Queue" to "Queue"

• Page 112, Figure 4.4, line 3: Change static to local variables.

• Page 113, line 4: Change "A realistic problems has surface that looks" to "Realistic problems often have surfaces that look"

Chapter 5

• Page 124, line -2: Change "A11" to "A1". Figure 5.2 is correct in saying that A1 is MAX's best move, and A11 is MIN's best reply.

• Page 128, 130: The outlines from the white chess pieces are missing. (What happened is that they were drawn with a line wdth that was about right for the printer we tested with, but on the high-resolution printer used for the final copy, they became invisible.) In Figure 5.4 there are supposed to be captured white pieces on the side, but they can't be seen.

Chapter 6

• Page 167 line 13: Change "an atomic sentences" to "an atomic sentence"

• Page 170 line -10: Change "the fewer the models" to "the fewer models".

Chapter 7

• Page 196, Subsection "The uniqueness operator i", line 7: Change "Ruler(r,Freedonia) ^ Dead(r)" to "Ruler(r,Freedonia) ^ Forall s Ruler(s,Freedonia) => Dead(s)"
• Page 201 line 8: Change "queries or goal" to "queries or goals"

• Page 205, Figure 7.3: Between S1 and S3 should be S2. It currently appears to have a 2 superimposed on a 0.

• Page 208, line 27: Change "a = Turn(Right) AND a = Turn(Left)" to "a = Turn(Right) OR a = Turn(Left)" (using the V-shaped symbol for OR)

• Page 210, line 17: Change "an OK action" to "a medium action"
  Also line 22: Change "OK(b, s)" to "Medium(b, s)"
  (Note that OK is a predicate on squares, while Medium is a predicate on actions.)

• Page 216, Exercise 7.16b: Change "wumpus world" to "wumpus worlds"

Chapter 8

• Page 239 lines 11-12: Seconds is in the wrong place. Change to:

  Time(Start(AD1991)) = Date(00,00,00,Jan,1,1991)
  Date(12,34,56,Feb,14,1993) = Seconds(2938682096)

• Page 245 line 16: Change "or or" to "or"

• Page 245 line -10: Change "[``=>'']" to "``=>''"

Chapter 9

• Page 275: end of line 8: Add "(Here UNIFY is assumed to standardize apart its second argument.)"

• Page 276: Figure 9.3: Remove the node HighTech(y). The nodes to the right should be Weapon(y) and then Missle(y) below, with the substitution {y/M1}, just as in Figure 9.4.
  Also, add "Yes" to the substitutions in Figure 9.4.

Chapter 11

• Page 339, lines 1: Change "descriptions," to "descriptions."

• Page 339, line -2: Change "and goals and goals" to "and goals"

• Page 353, line 17: Change "backs up to a try" to "backs up to try"

• Page 354, first bullet: Change "go(HWS)" to "Go(HWS)"

• Page 357, line -3: Change "that the each" to "that each"

• Page 362, second bullet: Change "that can" to "can"

Chapter 13

• Page 404, line 18: Change "/Clear(y)" to "~Clear(y)" (where ~ represents the NOT operator)

Chapter 15

• Page 443, line -10: Change "the the inflow" to "the inflow"

• Page 448, line 9: Change "PX|E" to "P(X|E)"

• Page 451, line -6: Change "then our recursions would either count the same evidence more than once or they would fail to terminate" to "then either our recursions would count the same evidence more than once or they would fail to terminate."

Chapter 16

• Page 472, line 1 of paragraph 2: Change "An nondeterministic" to "A nondeterministic"
  Also change "i" to "the index i" and in the next line change "Prior to execution" to "Prior to the execution"

• Page 484, line 2 of section 16.5: Change "a influence" to "an influence"

Chapter 17

• Page 501, line -1: After "complex ways" add "that depend on its current beliefs"

• Page 511, Figure 17.9, second assignment statement: "P Xt" should be "Xt" (as in equation 17.9, page 510).

Chapter 18

• Page 540, line 1: Change "Quinlan 1993" to "Quinlan, 1993a"

• Page 544, just before 18.5: Change "packages been" to "packages have been"

• Page 556, line 16: Change "examples, It" to "examples. It"

• Page 559, line 15: Change "Quinlan 1993" to "Quinlan, 1993a"

Chapter 19

• Page 588, first equation: missing a summation sign (over i) on the right-hand side.

• Page 588, second equation: delete "P(Hmap|D)"

Chapter 20

• Page 600, Section 20.2, line 7: Change "(3,2) or (3,3)" "(4,2) or (4,3)". The training sequences also need to be changed so that they all end in (4,2) or (4,3). (We originally used a 3x3 world for the example, and then switched to a 4x3 world..)

• Page 601, Figure 20.1: Change the entry for cell (2,1) to "-0.4177"

• Page 602, Figure 20.3: The for loop should be indented, making it part of the then part of the conditional. In other words, the loop is only executed when the percept is a terminal state.

• Page 616, line -14: Change "Quinlan 1993" to "Quinlan, 1993b"

Chapter 21

• Page 626, line -4: Change "Thag" to "Zog" (We had mis-remembered the cartoon.)

• Page 629, line -3: Change "Thag" to "Zog"

• Page 643, Figure 21.8, line 6: add target as second argument to the function call to New-Clause.
  Also, in function Extend-Example, change "example, literal" to "example, literal" (variables should be in italics)

• Page 646, line 9: Change "analogy reasoning" to "analogical reasoning".
  Also, line 16: Change the sentence "Dependencies were ... 1987))." to Dependencies were independently rediscovered, and given a full logical analysis, by Davies and Russell (Davies, 1985; Davies and Russell 1987) for the problem of analogical inference."
  Also, in footnote, change "1986b" to "1986".

Chapter 22

• Page 652, line 1 of section 22.1: Change "One of the actions that is available" to "One of the actions available" (thereby bypassing the question of "is" versus "are".)

• Page 654, last sentence of section 22.1: Change "Natural language understanding is one of the few areas of AI that has this property" to "Natural language understanding is one of the few areas of AI that have this property"

• Page 656, line 10: Change "the right-hand hand side" to "the right-hand side"

• Page 658, line 7: Change "the the Latin phrase" to "the Latin phrase"
  line 11: Change "nodes represents" to "nodes represent"
  lines 12: Change "We define" to "If we define"

• Page 664, last sentence of third paragraph: another incorrect parallel structure; change "on the decline" to "were on the decline"

• Page 667, line -4: Change "predicates" to "predicate"

Chapter 24

• Page 756, Figure 24.30: Change "AD.BC" to "AC.BD" and change "A'D'.B'C'" to "A'C'.B'D'".

Chapter 25

• Page 789, line 2: change "proposition" to "propositions"

• Page 790, Figure 25.10: change to a larger font. Also, in footnote 6, add quotes around "state space", and add "the" before "robotics and control literature".

Chapter 26

• Page 818, lines 12-13: Change "philosophical questions concerning which the interests of AI and philosophy coincide" to "those areas of philosophical investigation in which AI and philosophy are on the same team, so to speak"

• Page 818, line 16: Change "intelligent." to "intelligent?"

• Page 818, line -13: Delete "cannot" from "machines that act intelligently cannot have real, conscious minds."

  (To clear things up, the strong AI position defines "intelligence" as something that requires real, conscious, human-like minds. So the position is really one of how to define the field, not of what side one takes. However, most people who accept the strong AI position are philosophers who are claiming that computers do not achieve this (regardless of the seemingly intelligent behavior they display in some tasks) and thus they are claiming that strong AI does not (or sometimes can not) exist. Less often, people claim that strong AI does exist. )

Chapter 27

• Page 843, line -12: Change "(Hanks et al., 1994)" to "(Hanks et al., 1994)"

Appendix B

• Page 854, line 9: Change "English)" to "English"

• Page 857, line -6: After "07632." add "Finally, a good deal of information about the book, including the code, can be obtained from the World-Wide Web at the URL

        http://www.cs.berkeley.edu/~russell/aima.html
  

Bibliography

• page 859, column 1, last line: "la risque" should be "le risque"

• column 2, first line: "Ame'ricaine" should be "ame'ricaine" (as an adjective, the word is not capitalized in French)

• page 866, column 1, line 14: "metamorphose" should be "me'tamorphose" (acute accent missing)

• line 30: "syste'me" should be "syst`eme" (wrong accent)

• line 31: "Franc,ais" should be "franc,ais" (the noun designating the language is not capitalized)

• page 867, column 2, line 29: "Le pre'vision" should be "La pre'vision"

• Page 891: Changed Quinlan (1993) to Quinlan (1993b).
  Added "Quninlan, J. R. (1993a). C4.5: Programs for Machine Learning. Morgan Kaufmann, San Mateo, California.

• page 893, column 1, line 8: "manual" should be "manuel"

• line 9: "reference" should be "re'fe'rence" (two acute accents missing)

• "utilization" should be "utilisation"

• Drop Russell (1986a) (it was referenced in error) and change Russell (1986b) to Russell (1986).

Index

• Page 924: Change "POMPD" to "POMDP"

• Page 930: Add "TD-Gammon, 617".
  Also, drop "Thag"

• Add "Zog, 626, 629"

Back Flap

• Stuart's biography needs updating: after "Presidential Young Investigator Award", add ", and in 1995 he received the Computers and Thought Award"

• Peter's affiliation and research project has changed. Change "Berkeley. He is currently a Staff Scientist at Sun Microsystems Laboratories." to "Berkeley, and Sun Microsystems Laboratories. He is currently a Computer Scientist at Harlequin Inc." Change the sentence that starts "His current research ..." to "His current research involves applications of decision theory, machine learning, information retrieval, and natural language processing to real-world problems."