Q: Expert Ease, an early expert-system tool/shell, was developed under the direction of which artificial intelligence pioneer?

Introduction / Context: Many notable AI systems and shells were shepherded by leading figures in the field. “Expert Ease” is remembered as an early expert-system tool associated with British AI efforts. This question tests historical knowledge: connecting a specific tool to the researcher who led its development. Given Data / Assumptions: We are identifying leadership (“under the direction of”). Candidate names include foundational AI pioneers and theorists. Expert Ease refers to an expert-system environment/tool. Concept / Approach: Donald Michie was a prominent British AI researcher who helped establish and direct AI research institutions and projects, including work on knowledge-based systems and tools. Expert Ease has been historically attributed to efforts under his direction. The other names, while seminal in AI, are associated with different milestones: John McCarthy (Lisp, time-sharing AI labs), Lotfi Zadeh (fuzzy logic), Alan Turing (foundations of computation). Step-by-Step Solution: Identify Expert Ease as an expert-system shell/tool.Recall British AI leadership: Donald Michie’s direction over such projects.Eliminate other pioneers whose signature contributions lie elsewhere.Choose “Donald Michie.” Verification / Alternative check: Historical AI literature and retrospectives on UK expert systems reference Michie’s leadership in expert-system development programs, aligning Expert Ease with his direction rather than with the other figures listed. Why Other Options Are Wrong: John McCarthy: Lisp and foundational AI, but not the director tied to Expert Ease. Lotfi Zadeh: fuzzy logic originator, not Expert Ease. Alan Turing: foundational theorist; predates expert-system shells. None of the above: incorrect because Michie is correct. Common Pitfalls: Associating any expert-system tool with McCarthy due to Lisp; overlooking UK-centric AI developments led by Michie in the era of knowledge-based systems. Final Answer: Donald Michie

Q: MLMenu, a natural-language interface for the TI Explorer system, is most similar to which named natural-language interaction technology?

Introduction / Context: Natural-language (NL) interfaces allow users to interact with systems using phrases or sentences, often mapping them to commands or database queries. MLMenu was a natural-language interface available on the TI Explorer, a Lisp machine platform. This question asks you to identify the most similar named technology among the options given. Given Data / Assumptions: MLMenu is an NL interface (dialog/menu-driven with NL flavor). We seek similarity to another NL interaction technology by name. Non-NL technologies (networking hardware, languages, shells) should be excluded. Concept / Approach: NaturalLink is an example name associated with natural-language interfaces that map user phrases to system actions. By contrast, Ethernet is a networking standard; PROLOG is a logic programming language; and the Personal Consultant refers to an expert-system shell, not specifically an NL interface. Thus, an NL-to-NL analogy aligns MLMenu most clearly with NaturalLink. Step-by-Step Solution: Categorize each option as NL interface vs. non-NL technology.Match MLMenu (NL interface) to the analogous NL interface name: NaturalLink.Eliminate unrelated items (networking, languages, shells).Select “NaturalLink.” Verification / Alternative check: Historical references to TI Explorer ecosystems include NL and menu-driven front ends akin to NaturalLink, reinforcing the similarity classification rather than the unrelated technologies listed. Why Other Options Are Wrong: Ethernet: physical/data-link networking; unrelated to NL interaction. PROLOG: programming language; not specifically an NL interface. the Personal Consultant: expert-system shell; does not denote an NL interface. None of the above: incorrect because NaturalLink is a relevant analogue. Common Pitfalls: Confusing knowledge-based shells with NL front ends; assuming any AI-related term like PROLOG implies an NL interface by itself. Final Answer: NaturalLink

Question 1

In the history of artificial intelligence, A. M. Turing proposed a practical technique to decide whether a computing machine could convincingly demonstrate human-like intelligence in conversation. Today, this famous conversational criterion is known…

Accepted Answer

Introduction / Context: Early artificial intelligence research needed a way to talk concretely about whether machines could be said to “think.” Alan Mathison Turing, a foundational figure in computer science, proposed an operational criterion rather than a philosophical debate: if a machine could hold a text-based conversation that a human evaluator could not reliably distinguish from a human interlocutor, the machine could be considered intelligent for practical purposes. This idea became widely known as the Turing Test. Given Data / Assumptions: The question asks for the present-day name of Turing’s proposed conversational criterion. It concerns human–machine dialogue and indistinguishability from human responses. The focus is on a practical test, not on mathematical systems or general algorithms. Concept / Approach: The Turing Test frames intelligence as observable behavior in conversation. The evaluator interacts via text-only channels to avoid bias from appearance or voice. If the evaluator cannot reliably tell which participant is the machine, the machine is said to have passed the test in that setting. This avoids metaphysical questions and emphasizes measurable performance—a theme that influences many modern AI benchmarks. Step-by-Step Solution: Identify the historical figure and contribution: Alan Turing proposed a conversational test for intelligence. Map the proposal to its modern name: this criterion is called the Turing Test. Eliminate distractors that are mathematical (Boolean algebra, logarithm) or generic (algorithm) rather than a named test. Select the answer: Turing Test. Verification / Alternative check: Decades of AI history cite the Imitation Game (1950) as the origin of the Turing Test. Contemporary discussions of AI capability often refer back to this behavioral standard, even as more specialized benchmarks have proliferated. Why Other Options Are Wrong: Boolean Algebra: a mathematical logic system, not a conversational criterion. Algorithm: a general procedure; not Turing’s named test. Logarithm: a mathematical function, unrelated to AI evaluation. None of the above: incorrect because the correct name is the Turing Test. Common Pitfalls: Confusing the Turing Test with general problem-solving ability or assuming it involves speech, vision, or robotics; the original proposal used text-only interaction. Final Answer: Turing Test

Question 2

In automatic speech recognition, which approach most effectively avoids errors due to coarticulation and context-dependent pronunciation variations by requiring pauses between words?

Accepted Answer

Introduction / Context: Human speech exhibits coarticulation—sounds influence each other depending on their neighbors, speaking rate, and prosody. This makes automatic speech recognition difficult because the same word can sound different in different contexts. Recognition strategies differ in how they handle such variability and the timing between words. Given Data / Assumptions: The goal is to avoid errors caused by context-dependent pronunciation and coarticulation. We compare isolated, connected, and continuous recognition modes (and a speaker dependency setting). We assume a standard microphone input and typical acoustic models. Concept / Approach: Isolated word recognition requires the speaker to pause between each word. The pause allows the system to segment the audio cleanly and model each word without strong coarticulation influences from adjacent words. By contrast, connected word recognition permits short pauses and retains some coarticulation, and continuous speech recognition attempts to model fully fluent speech, which maximizes coarticulation and contextual variation. Speaker-dependent systems are trained on one speaker but still face coarticulation in connected or continuous modes. Step-by-Step Solution: Identify the source of difficulty: coarticulation between adjacent words. Select the approach that enforces clear boundaries: isolated word recognition with pauses. Eliminate approaches that retain coarticulation (connected, continuous). Note that speaker dependence is orthogonal to segmentation and does not eliminate coarticulation. Verification / Alternative check: Classical systems and early voice-command products often required the user to speak one word at a time specifically to simplify segmentation and reduce coarticulation effects, validating the choice. Why Other Options Are Wrong: Connected: still has reduced but present coarticulation. Continuous: has the most coarticulation and contextual variability. Speaker-dependent: does not inherently remove coarticulation. None: incorrect because isolated word recognition is the precise fit. Common Pitfalls: Assuming speaker-dependent systems eliminate all variability; they personalize models but still contend with continuous-speech dynamics if used in that mode. Final Answer: Isolated word recognition

Question 3

Historical foundations of AI: which Stanford University professor coined the term “artificial intelligence” in 1956 at the Dartmouth Conference?

Accepted Answer

Introduction / Context: The Dartmouth Summer Research Project on Artificial Intelligence (1956) is widely considered the birth of AI as a formal academic discipline. The proposal and gathering set the agenda for decades, framing AI as the study of making machines perform tasks that would require intelligence if done by humans. Given Data / Assumptions: Date and event: 1956, Dartmouth Conference. Task: identify the Stanford professor who coined “artificial intelligence.” Several prominent computing figures are listed as distractors. Concept / Approach: John McCarthy, then at Dartmouth and later at Stanford, coined the term “artificial intelligence” and co-organized the conference with Marvin Minsky, Nathaniel Rochester, and Claude Shannon. He later contributed LISP and time-sharing concepts, cementing his influence on AI’s trajectory. Step-by-Step Solution: Recall the Dartmouth Conference organizers and coinage of the term “AI.” Match the name to the contribution: John McCarthy → term originator. Eliminate other figures (e.g., Weizenbaum wrote ELIZA; others worked in chess or different areas). Select John McCarthy as the correct answer. Verification / Alternative check: AI histories consistently cite McCarthy as the term’s originator. Stanford CS and AI Lab retrospectives also attribute the coinage to him. Why Other Options Are Wrong: Joseph Weizenbaum: created ELIZA (1960s), not the AI term. David Levy and Hans Berliner: prominent in computer chess, not in naming AI. None of the above: incorrect because McCarthy is correct. Common Pitfalls: Confusing AI pioneers’ distinct contributions; mixing the Dartmouth organizers with other contemporaries. Final Answer: John McCarthy

Question 4

Early AI systems: which program, developed by Daniel G. Bobrow, read and solved algebra word problems by interpreting natural-language statements into equations?

Accepted Answer

Introduction / Context: Classic AI included landmark systems that demonstrated specific capabilities. One early achievement was parsing plain-language algebra problems and computing solutions—bridging natural language understanding and symbolic mathematics. Daniel G. Bobrow’s system is often cited in AI histories for this feat. Given Data / Assumptions: The system read algebra word problems. It translated natural language into formal equations and solved them. We compare well-known AI names from the same era. Concept / Approach: Bobrow’s program named STUDENT interpreted textual algebra problems into simultaneous equations and solved them, demonstrating early semantic parsing. SHRDLU (Winograd) worked in the Blocks World for dialog and manipulation. BACON (Langley) explored scientific discovery heuristics. SIMD is a hardware parallelism acronym, not an AI program title. Step-by-Step Solution: Identify the target capability: algebra word-problem solving from text. Associate capability with the system name → STUDENT. Eliminate other programs with different scopes (SHRDLU, BACON) or non-program acronyms (SIMD). Select STUDENT as correct. Verification / Alternative check: AI textbooks and retrospectives consistently attribute algebra word-problem parsing to Bobrow’s STUDENT, validating the identification. Why Other Options Are Wrong: SHRDLU: natural-language dialog in a microworld, not algebra text problems. BACON: scientific discovery heuristics, not algebra word problems. SIMD: Single Instruction, Multiple Data—an architecture class, not the system in question. Common Pitfalls: Assuming any famous early NLP system solved algebra word problems; SHRDLU is often misattributed. Final Answer: STUDENT

Question 5

The “Turing Machine” result in computability theory showed that any algorithmic procedure can be programmed using a very simple formal system; in popular summaries this is often described as using a _____ representation to express any computable tas…

Accepted Answer

Introduction / Context: Turing machines formalize the notion of computation with a minimal set of operations on symbols written on an infinite tape. The central idea is that complex algorithms can be reduced to simple symbolic manipulations and still retain full computational power. In non-technical summaries, this universality is often illustrated by showing that two symbols (e.g., 0 and 1) are sufficient to express any computable procedure. Given Data / Assumptions: The question seeks the commonly stated minimal representation used to encode any algorithmic task. We assume the popular binary encoding perspective of computability results. Distractors include unrelated physical or linguistic characterizations. Concept / Approach: While Turing’s formalism does not require specifically binary symbols (any finite alphabet suffices), it established that a very small alphabet—even two symbols—is adequate. Hence, in popular computing lore, this becomes the claim that “binary” representation can encode any computation, aligned with modern digital computers using bits. Step-by-Step Solution: Recognize the universality of Turing machines and coding of algorithms as symbol strings. Recall that two symbols suffice to represent arbitrary computations. Map this to the common term used in computing practice → binary. Select “binary” as the widely accepted answer. Verification / Alternative check: Church–Turing thesis discussions and CS curricula highlight that binary encoding of instructions and data is sufficient and standard in digital computers, reflecting Turing’s universality insight. Why Other Options Are Wrong: Electro-chemical: a physical substrate, not a symbolic representation. Recursive: a property of functions/definitions; not the minimal code alphabet. Semantic: relates to meaning, not representation alphabet. None: incorrect because binary captures the intended idea. Common Pitfalls: Conflating the formal statement (any finite alphabet) with the practical convention (binary); for exam purposes, binary is the expected choice. Final Answer: binary

Question 6

In philosophical classifications of AI, “Strong AI” refers to which position about machine minds and cognition?

Accepted Answer

Introduction / Context: Artificial intelligence discussions often distinguish between “weak” (or “narrow”) AI and “strong” AI. The distinction is philosophical as much as technical and centers on whether a machine can possess a mind or genuine understanding, as opposed to simply simulating intelligent behavior. Given Data / Assumptions: The question targets the meaning of “Strong AI.” Options include weaker positions about simulation or study of mind. We assume the standard philosophical usage. Concept / Approach: Strong AI asserts that suitably programmed computers can have real minds, consciousness, or understanding—not just the appearance of intelligence. In contrast, claims about producing intelligent-looking outputs (option b) or using computers to study cognition (option c) align with weak AI or cognitive modeling perspectives, not Strong AI. Step-by-Step Solution: Identify the definition that attributes genuine mentality to machines. Option (a) matches the Strong AI thesis. Reject options describing mere simulation (b) or modeling for study (c). Therefore, select (a). Verification / Alternative check: Philosophy of mind and AI textbooks consistently define Strong AI as the thesis that a computer can literally understand and have mental states if appropriately programmed. Why Other Options Are Wrong: (b) Describes weak AI: simulation without claim of understanding. (c) Describes computational cognitive science, not Strong AI. (d) “All” is false because the three positions conflict. (e) “None” is false because (a) is correct. Common Pitfalls: Confusing success in tasks with claims about consciousness; equating performance benchmarks with philosophical commitments. Final Answer: The embodiment of human intellectual capabilities within a computer (machines can genuinely think).

Question 7

In Common LISP programming, which built-in predicate returns t if is even and nil otherwise (assume is an exact integer object)?

Accepted Answer

Introduction / Context: Common LISP provides a rich set of numeric predicates for testing properties of numbers. Knowing the idiomatic names of these predicates is important for writing readable, conventional LISP code and for understanding legacy AI codebases that rely on LISP macros and functions. Given Data / Assumptions: Language: Common LISP. Input is an exact integer. We need the standard predicate that tests parity. Concept / Approach: The canonical predicates for parity in Common LISP are evenp and oddp. The suffix “p” conventionally marks a predicate returning t or nil. There is no standard predicate named even, and fabricated names like numeven or numnevenp are not part of the standard. Step-by-Step Solution: Recall Common LISP naming conventions: predicates often end with “p”. Identify parity test: evenp → returns t for even integers, nil otherwise. Eliminate non-standard or misspelled alternatives. Choose (evenp ). Verification / Alternative check: Common LISP HyperSpec documents evenp and oddp as standard numeric predicates for integers, confirming the choice. Why Other Options Are Wrong: (even ): not a standard predicate. (numeven ), (numnevenp ): invented names. None: incorrect because (evenp ) is correct. Common Pitfalls: Forgetting the “p” suffix for predicates in LISP; confusing Common LISP with Scheme function names. Final Answer: (evenp )

Question 8

Research programs exploring heuristic discovery: which series of AI systems by Pat Langley investigated how heuristics can drive scientific law discovery from data?

Accepted Answer

Introduction / Context: One line of AI research investigated how scientific laws might be induced from empirical data using heuristic search and representation choices. Such systems illuminated how concept formation, measurement selection, and search biases can uncover regularities that resemble scientific laws. Given Data / Assumptions: The question asks for the series name associated with Pat Langley. The goal of the series was to explore the role of heuristics in scientific discovery. Other options are distractors without direct relevance. Concept / Approach: Pat Langley and collaborators developed the BACON series, which modeled the discovery of quantitative laws (e.g., proportionalities) from data. The systems used heuristic strategies to guide search through hypothesis spaces, reflecting how practical scientific reasoning often proceeds. Step-by-Step Solution: Identify the researcher: Pat Langley. Recall the project name: BACON series of discovery systems. Eliminate generic or unrelated terms (MIT, DU) and ambiguous acronyms. Select BACON. Verification / Alternative check: AI histories and machine discovery literature consistently associate Langley with BACON as a canonical example of discovery systems. Why Other Options Are Wrong: MIT: an institution, not the system series. RAMD/DU: do not name Langley’s discovery systems. None: incorrect because BACON is correct. Common Pitfalls: Confusing SHRDLU/ELIZA (NLP) with discovery systems; overlooking that BACON addresses scientific induction tasks. Final Answer: BACON

Question 9

Relative strengths of humans and computers: historically, which tasks have computers done better than people, especially in early AI eras?

Accepted Answer

Introduction / Context: Comparing human and machine capabilities clarifies where AI brings leverage. Historically, computers excel at deterministic, high-volume, and precise tasks, while humans excel at flexible, context-rich reasoning. Early AI emphasized symbolic reasoning and search but relied on the computational strengths of storage and arithmetic provided by digital hardware. Given Data / Assumptions: We consider traditional, long-recognized strengths of computers. We contrast them with human strengths such as flexible adaptation. We focus on tasks rather than philosophical claims. Concept / Approach: Computers are unmatched at rapid, error-free arithmetic and at storing/retrieving large volumes of structured information. They struggle with open-ended novelty and contextual interpretation without extensive training or rules. Therefore, “storing information” and “computing numerically” align with machine strengths; “responding flexibly” characterizes human advantage, especially in earlier eras. Step-by-Step Solution: List machine strengths: high-speed arithmetic, memory/storage, exact repetition. List human strengths: creativity, generalization from few examples, commonsense. Match options accordingly and choose the combined strengths for computers. Select (d): both storing information and computing numerically. Verification / Alternative check: Benchmarks from the earliest computers onward demonstrate that arithmetic throughput and data storage scale with hardware, while flexible cognition remains challenging; this is why AI research exists as a distinct field. Why Other Options Are Wrong: Responding flexibly: historically a human strength, not a traditional machine advantage. Only storage or only arithmetic: each is true, but the combined option captures both strengths. None: incorrect because computers excel at both (b) and (c). Common Pitfalls: Overgeneralizing recent machine-learning advances to all forms of flexibility; ignoring the historical context of the question. Final Answer: Both storing information and computing numerically

Question 10

In AI system design, the “output” stage can take multiple forms. Which of the following are valid output modalities or actions that an AI program might produce to affect the world or communicate results?

Accepted Answer

Introduction / Context: AI systems often involve a pipeline: perception or input, internal reasoning/planning, and outputs that communicate or act. Outputs are not limited to text on a screen; they may include spoken language, mechanical actuation, and navigation—especially in robotics and embodied AI. Given Data / Assumptions: The term “output” encompasses both communicative and physical actions. We consider common AI applications: conversational agents, assistive robots, autonomous vehicles. Any action that externalizes the system’s decisions qualifies as output. Concept / Approach: Output modalities include: printed language (reports, screens, printers) and synthesized speech for human-friendly communication; manipulation (robot arms, grippers) to exert forces and change the environment; and locomotion to change position in the world. These are standard in HRI (human–robot interaction), industrial automation, and assistive AI technologies. Step-by-Step Solution: Identify communication outputs: text/print and text-to-speech. Identify embodied outputs: manipulation via actuators. Identify navigation outputs: locomotion commands to motors/drive systems. Conclude that all listed modalities are valid outputs → choose “All of the above.” Verification / Alternative check: Practical systems—from dialogue agents with speech synthesis to factory robots and autonomous mobile robots—demonstrate these outputs daily, confirming the breadth of AI output channels. Why Other Options Are Wrong: Any single-modality answer is incomplete because AI outputs span multiple forms. None: plainly false; these are routine outputs in AI applications. Common Pitfalls: Thinking of AI as purely software with on-screen text; overlooking robotics and voice interfaces as integral output pathways. Final Answer: All of the above

Question 11

In SIMD (Single Instruction stream, Multiple Data stream) computer architecture, how many arithmetic logic units (ALUs) can be directed by a single control unit at the same time? Provide the most accurate characterization.

Accepted Answer

Introduction / Context: SIMD (Single Instruction stream, Multiple Data stream) is a classic parallel architecture classification in computer organization. In a SIMD machine, one control unit fetches and decodes a single instruction, and that instruction is executed simultaneously across multiple processing elements (PEs) or arithmetic logic units (ALUs) on different data items. This question checks your understanding of how many ALUs a single control unit can direct in such systems. Given Data / Assumptions: Architecture under discussion: SIMD. There is exactly one instruction stream and multiple data streams. The control unit broadcasts or otherwise dispatches the same instruction to many processing elements. Concept / Approach: In SIMD, the defining feature is one control unit driving multiple execution lanes. The number of ALUs is not fixed to a specific constant; it varies by design and scale. Historic examples include array processors and vector supercomputers with many ALUs; modern examples include GPUs or vector units (e.g., SSE/AVX/NEON) with lane counts that differ by generation. Therefore, the accurate characterization is that a single control unit can control one or more ALUs, typically many in practical designs. Step-by-Step Solution: Identify the architectural model: single control unit, multiple execution lanes.Recognize that ALU count is a design choice (scalable).Match wording: “one or more ALUs” captures scalability and correctness.Reject fixed counts that contradict scalability. Verification / Alternative check: Real systems (vector supercomputers, GPUs, and SIMD instruction sets in CPUs) demonstrate many ALUs executing the same instruction on packed data. Documentation consistently describes a broadcast control to multiple lanes, not a fixed two-ALU limit or single-ALU constraint. Why Other Options Are Wrong: One ALU: contradicts the “multiple data” premise. Only two ALUs: arbitrary limit; not inherent to SIMD. Impossible to say: we can say it is one-or-more, typically many. None of the above: invalid because one-or-more is correct. Common Pitfalls: Confusing SIMD with MIMD (multiple instruction streams) or SISD (scalar). Assuming a fixed ALU count rather than a scalable array or vector width that can grow across implementations. Final Answer: one or more ALUs

Question 12

In computer vision, which technique explicitly relies on image templates (stored exemplars) to recognize or locate objects in scenes?

Accepted Answer

Introduction / Context: Computer vision encompasses many approaches to understanding images, from low-level feature extraction to high-level object recognition. Some approaches match observed image data to previously stored exemplars or “templates.” This question asks you to identify which option describes a technique that relies on such templates to interpret a scene. Given Data / Assumptions: The technique should use templates or exemplars of object appearance/shape. We contrast it with methods that focus on stereo geometry, basic edges, or the broader application of vision to robotics. Concept / Approach: Model-based vision typically uses known object models (which can be geometric or appearance templates) to explain observed image evidence. The system compares features or image patches to stored templates of the target object to hypothesize presence, pose, or segmentation. Template matching is a classic component here. By contrast, edge detection is a low-level operation producing gradient-based edge maps; binocular vision (stereo) estimates depth via disparity; and robot vision is an application context, not a specific template-driven technique. Step-by-Step Solution: Identify the key phrase: “relies on image templates.”Map to concept: model-based vision matches observed data to stored object models/templates.Eliminate edge detection (feature extraction), binocular vision (geometry), and robot vision (application area).Select “model-based vision.” Verification / Alternative check: Standard texts describe template matching and model fitting (2D/3D) as canonical model-based methods for recognition and pose estimation, often using correlation, chamfer matching, or feature-based alignment to templates. Why Other Options Are Wrong: Edge detection: extracts gradients; does not use stored templates by itself. Binocular vision: computes disparities; template use is not central. Robot vision: context of use, not a particular technique. None of the above: incorrect because model-based vision fits. Common Pitfalls: Equating any recognition with “robot vision”; assuming edge maps themselves are templates rather than inputs for later matching stages. Final Answer: model-based vision

Question 13

In Common Lisp, evaluate the expression and predicate:  What is the result of (minusp (- 20 4 8 8 1)) ?

Accepted Answer

Introduction / Context: Common Lisp provides arithmetic operators and numeric predicates that return truth values. The function minusp tests whether a number is strictly less than zero. This question combines arithmetic evaluation with a predicate: you must first compute the numeric result of a subtraction expression and then apply minusp to that result. Given Data / Assumptions: Expression to evaluate is (- 20 4 8 8 1). Predicate to apply is (minusp ), which returns T if < 0, else NIL. In Common Lisp, subtraction with multiple arguments computes left-associative subtraction: a - b - c - d ... Concept / Approach: Compute the inner arithmetic first, then use the predicate. Remember that T is Lisp’s canonical true value and NIL is false. Some Lisps also print NIL for false, not “F”. Therefore, the final answer must respect Lisp’s true/false conventions. Step-by-Step Solution: (- 20 4 8 8 1) → start with 20.20 - 4 = 16.16 - 8 = 8.8 - 8 = 0.0 - 1 = -1.Now evaluate (minusp -1) → is -1 < 0? Yes.Thus, (minusp (- 20 4 8 8 1)) returns T. Verification / Alternative check: You can verify in a REPL: evaluating the arithmetic yields -1, and (minusp -1) returns T. Testing a non-negative like 0 would return NIL because 0 is not negative in Lisp’s numeric predicates. Why Other Options Are Wrong: F: not a standard Lisp boolean; false is NIL. NIL: would be correct only if the number were non-negative. -20: confuses result with the original operand; the arithmetic result is -1, not -20. None of the above: incorrect because T is correct. Common Pitfalls: Mis-evaluating multi-argument subtraction; assuming 0 is negative; confusing NIL with F; forgetting Lisp’s truth values and evaluation order (inner form first). Final Answer: T

Question 14

In Common Lisp, which form evaluates the value expression and assigns that value to the (unevaluated) symbol to create a named constant?

Accepted Answer

Introduction / Context: Common Lisp has several defining forms for variables and constants. When you need to bind a symbol to a value that should not change (i.e., a constant), you use the defconstant form. It takes the symbol name (not evaluated) and an initializing form (evaluated) whose resulting value becomes the constant’s value. Given Data / Assumptions: The symbol (name) should be treated as an unevaluated identifier. The value expression should be evaluated, and that value assigned. We are specifically creating a constant, not a special/dynamic variable. Concept / Approach: The form (defconstant name value-form) defines name as a global constant. The name is a symbol, not evaluated; the value-form is evaluated at definition time, and the result is stored. Afterward, redefining may be implementation-dependent or produce warnings because constants are meant to be stable. Other forms in the options are either non-existent in Common Lisp or do not match the semantics described. Step-by-Step Solution: Identify the required behavior: evaluate the value form, not the symbol.Map to Common Lisp defining forms: defconstant, defparameter, defvar.Select defconstant because the question explicitly references an “unevaluated symbol” receiving the evaluated value and being a constant.Eliminate non-standard or incorrect forms. Verification / Alternative check: In a Lisp REPL, (defconstant +pi+ 3.14159) defines a constant. The symbol +pi+ is not evaluated as a variable at read time; it is used as a name. The numeric literal is evaluated to itself and stored as the constant value. Why Other Options Are Wrong: (constant ...): not a standard defining form. (eva ...): not a Common Lisp form; “eval” is different and not used like this. None of the above: incorrect because defconstant matches exactly. Common Pitfalls: Confusing defconstant with defparameter (which allows reassignment) or defvar (which initializes only if unbound). Misunderstanding evaluation: symbol names in defining forms are not evaluated. Final Answer: (defconstant

Question 15

Expert Ease, an early expert-system tool/shell, was developed under the direction of which artificial intelligence pioneer?

Accepted Answer

Introduction / Context: Many notable AI systems and shells were shepherded by leading figures in the field. “Expert Ease” is remembered as an early expert-system tool associated with British AI efforts. This question tests historical knowledge: connecting a specific tool to the researcher who led its development. Given Data / Assumptions: We are identifying leadership (“under the direction of”). Candidate names include foundational AI pioneers and theorists. Expert Ease refers to an expert-system environment/tool. Concept / Approach: Donald Michie was a prominent British AI researcher who helped establish and direct AI research institutions and projects, including work on knowledge-based systems and tools. Expert Ease has been historically attributed to efforts under his direction. The other names, while seminal in AI, are associated with different milestones: John McCarthy (Lisp, time-sharing AI labs), Lotfi Zadeh (fuzzy logic), Alan Turing (foundations of computation). Step-by-Step Solution: Identify Expert Ease as an expert-system shell/tool.Recall British AI leadership: Donald Michie’s direction over such projects.Eliminate other pioneers whose signature contributions lie elsewhere.Choose “Donald Michie.” Verification / Alternative check: Historical AI literature and retrospectives on UK expert systems reference Michie’s leadership in expert-system development programs, aligning Expert Ease with his direction rather than with the other figures listed. Why Other Options Are Wrong: John McCarthy: Lisp and foundational AI, but not the director tied to Expert Ease. Lotfi Zadeh: fuzzy logic originator, not Expert Ease. Alan Turing: foundational theorist; predates expert-system shells. None of the above: incorrect because Michie is correct. Common Pitfalls: Associating any expert-system tool with McCarthy due to Lisp; overlooking UK-centric AI developments led by Michie in the era of knowledge-based systems. Final Answer: Donald Michie

Question 16

In manufacturing, which discipline explicitly relates individual stages to one another and coordinates them as an integrated whole across the project lifecycle?

Accepted Answer

Introduction / Context: Manufacturing involves multiple stages—requirements, design, prototyping, production, quality assurance, distribution, and field service. A key management discipline ensures that all stages align with schedule, budget, scope, and quality goals, and that inter-stage dependencies are coordinated. This question asks which role relates each stage to every other and to the process as a whole. Given Data / Assumptions: We are considering oversight across all phases of manufacturing. The answer should involve planning, coordination, and control of resources and timelines. Other options are specific functions (e.g., design, distribution) rather than cross-cutting governance. Concept / Approach: Project management is the discipline that integrates scope, time, cost, quality, communications, risk, procurement, and stakeholders across the entire lifecycle. It connects early design decisions to downstream manufacturing and distribution, synchronizes schedules, and manages change so that the “whole” process remains coherent. The other options represent important but localized functions that do not span the entire lifecycle by mandate. Step-by-Step Solution: Identify the need: an integrative function across all stages.Map manufacturing stages to a lifecycle; note interdependencies.Recognize that project management explicitly coordinates these dependencies and stages.Select “project management.” Verification / Alternative check: Standard bodies of knowledge (e.g., PMI’s PMBOK) define integration management and cross-phase coordination as primary responsibilities of project management, confirming the choice. Why Other Options Are Wrong: Field service: post-delivery support; not integrative across all stages. Design: one stage; does not manage the entire lifecycle. Distribution: outbound logistics; a later-stage function. None of the above: incorrect because project management matches the description. Common Pitfalls: Equating manufacturing operations with project management; overlooking that project managers orchestrate cross-functional teams to deliver the product through all stages. Final Answer: project management

Question 17

MLMenu, a natural-language interface for the TI Explorer system, is most similar to which named natural-language interaction technology?

Accepted Answer

Introduction / Context: Natural-language (NL) interfaces allow users to interact with systems using phrases or sentences, often mapping them to commands or database queries. MLMenu was a natural-language interface available on the TI Explorer, a Lisp machine platform. This question asks you to identify the most similar named technology among the options given. Given Data / Assumptions: MLMenu is an NL interface (dialog/menu-driven with NL flavor). We seek similarity to another NL interaction technology by name. Non-NL technologies (networking hardware, languages, shells) should be excluded. Concept / Approach: NaturalLink is an example name associated with natural-language interfaces that map user phrases to system actions. By contrast, Ethernet is a networking standard; PROLOG is a logic programming language; and the Personal Consultant refers to an expert-system shell, not specifically an NL interface. Thus, an NL-to-NL analogy aligns MLMenu most clearly with NaturalLink. Step-by-Step Solution: Categorize each option as NL interface vs. non-NL technology.Match MLMenu (NL interface) to the analogous NL interface name: NaturalLink.Eliminate unrelated items (networking, languages, shells).Select “NaturalLink.” Verification / Alternative check: Historical references to TI Explorer ecosystems include NL and menu-driven front ends akin to NaturalLink, reinforcing the similarity classification rather than the unrelated technologies listed. Why Other Options Are Wrong: Ethernet: physical/data-link networking; unrelated to NL interaction. PROLOG: programming language; not specifically an NL interface. the Personal Consultant: expert-system shell; does not denote an NL interface. None of the above: incorrect because NaturalLink is a relevant analogue. Common Pitfalls: Confusing knowledge-based shells with NL front ends; assuming any AI-related term like PROLOG implies an NL interface by itself. Final Answer: NaturalLink

Question 18

Which of the following institutions was NOT among the earliest leaders in artificial intelligence research?

Accepted Answer

Introduction / Context: The origin of academic AI is linked to a small set of institutions that hosted seminal conferences, labs, and researchers. Knowing who the early leaders were helps contextualize the field’s development from the 1950s onward. This question asks you to identify which listed school was not among the earliest leaders. Given Data / Assumptions: “Early leaders” refers to the 1950s–1960s formative period. Institutions like MIT and Stanford are historically recognized AI hubs. Dartmouth hosted the famed 1956 AI workshop (Dartmouth Conference). Concept / Approach: Dartmouth College (often informally referenced) hosted the 1956 workshop where the term “Artificial Intelligence” was popularized. MIT and Stanford established world-leading AI labs (e.g., MIT AI Lab; Stanford’s AI Lab and later HAI). Harvard, while an elite university, was not typically cited as an early AI leader compared to these institutions and later CMU (also not listed here). Therefore, the correct choice for “not among the early leaders” is Harvard University. Step-by-Step Solution: List historically noted early AI hubs: Dartmouth (conference), MIT (AI Lab), Stanford (SAIL).Assess Harvard’s role in the same period: less central leadership in early AI.Select “Harvard University” as the outlier.Confirm that “None of the above” is invalid because one outlier exists. Verification / Alternative check: Histories of AI commonly emphasize Dartmouth, MIT, Stanford, and CMU as earliest leaders. Harvard’s stronger AI prominence emerged later and in more specific subareas, not as a foundational early hub. Why Other Options Are Wrong: Dartmouth University: hosted the pivotal 1956 event. MIT: home to an early, influential AI lab. Stanford: site of the famous SAIL and major expert-system work. None of the above: incorrect because Harvard is the outlier. Common Pitfalls: Confusing the Dartmouth conference with a department; overlooking CMU’s role (not listed here) when gauging “early leaders.” Final Answer: Harvard University

Question 19

In Common Lisp, what does the function (copy-list <list>) return?

Accepted Answer

Introduction / Context: In Common Lisp, lists are typically built from cons cells. When you copy a list, it is vital to understand whether the copy is shallow (top-level cons cells duplicated, elements shared) or deep (recursively duplicated). The standard function copy-list performs a shallow copy of a list’s top-level structure. Given Data / Assumptions: The function under discussion is copy-list. Argument is a proper list . We need to know exactly what the returned value represents. Concept / Approach: copy-list constructs a new chain of cons cells replicating the top-level structure of the input list while reusing (sharing) the element objects themselves. If the elements are atoms (numbers, symbols) or if they are lists, the inner lists are not duplicated—only their references are copied. Therefore, modifications to the new list’s top-level cons cells do not affect the original’s structure, but mutations to shared sublists or objects will be visible from both lists. Step-by-Step Solution: Call (copy-list L) on list L.Implementation allocates new cons cells for each top-level pair in L.It sets each new cons cell’s car to the original element reference and cdr to the next new cons cell.Result is a new list equal (equalp as appropriate) to L at top level. Verification / Alternative check: A common demonstration: let L = (list (list 1 2) 3). After (setf L2 (copy-list L)), altering (cdr L2) does not change L structure, but modifying the inner sublist (setf (caar L2) 9) affects both, because the sublist is shared. Why Other Options Are Wrong: Returns the length: that is length, not copy-list. Returns T if empty: that is essentially testing with null, not copying. All of the above: cannot be true because the other statements are false. None of the above: incorrect because the first statement is correct. Common Pitfalls: Confusing shallow vs. deep copy; expecting copy-list to duplicate nested lists recursively (use copy-tree for that). Final Answer: returns a new list that is equal to by copying the top-level elements of

Question 20

Which AI system, created by Terry A. Winograd, enabled interactive natural-language dialogue about a simplified “blocks world” domain?

Accepted Answer

Introduction / Context: Early natural-language understanding systems often restricted their domains to make language interpretation tractable. The “blocks world” is a simplified micro-world of geometric objects with actions like move, stack, and unstack. Terry A. Winograd’s system is famous for engaging in dialogue about this world, interpreting commands and answering questions in plain language. Given Data / Assumptions: The system supported interactive dialogue. The domain is the simplified blocks world. We need the specific system’s name among classic AI programs. Concept / Approach: SHRDLU is Winograd’s seminal natural-language system that parsed commands, reasoned about a blocks-world state, and carried out actions in a simulated environment. It integrated parsing, semantic interpretation, discourse context, and planning, illustrating how domain restriction could yield robust NL interaction. The other listed names correspond to different systems or concepts: SIMD is a parallel architecture class, BACON is a scientific discovery program (Langley et al.), and STUDENT (Weizenbaum’s ELIZA variants and other early programs) are not Winograd’s blocks-world system. Step-by-Step Solution: Identify the domain (blocks world) and interactive dialogue capability.Recall the classic system by Winograd that meets both criteria.Match name: “SHRDLU.”Eliminate distractors unrelated to Winograd or blocks world. Verification / Alternative check: Historical AI texts and videos demonstrate SHRDLU manipulating a virtual block scene, answering questions about the scene, and executing multi-step commands, affirming the identification. Why Other Options Are Wrong: SIMD: architecture category, not an NL system. BACON: program for scientific law discovery, not blocks world. STUDENT: different early AI program; not Winograd’s blocks world work. None of the above: incorrect because SHRDLU is correct. Common Pitfalls: Confusing SHRDLU with ELIZA or PARRY; mixing up program names from the same era. Final Answer: SHRDLU

Artificial Intelligence Questions