Download Generative AI Specialist Exam and more Exams Technology in PDF only on Docsity!
Question 1 Which of the following best describes Artificial Intelligence (AI)? A) A set of rules to manually process data B) Machines performing tasks that typically require human intelligence C) Robots used exclusively in industrial manufacturing D) A simple programming technique for automating calculations Correct Answer: B Explanation: AI involves creating machines or systems capable of performing tasks that usually require human intelligence, such as pattern recognition or decision-making. Question 2 Which milestone is considered an early breakthrough in AI research? A) The invention of the abacus B) The development of expert systems C) The creation of vacuum tubes D) The widespread use of email Correct Answer: B Explanation: Expert systems in the 1970s and 1980s were one of the first highly successful AI applications, showing that machines could mimic the decision-making of human experts.
Question 3 Which type of AI focuses on specific tasks rather than general intelligence? A) Superintelligence B) Narrow AI C) General AI D) Conscious AI Correct Answer: B Explanation: Narrow AI is designed to handle one specific task, such as image recognition or language translation, rather than having generalized human-like intelligence. Question 4 What defines General AI? A) AI capable of learning any intellectual task that a human can do B) AI limited to speech recognition C) AI that can only operate under supervision D) AI used only for automated email responses Correct Answer: A Explanation: General AI refers to machines with the ability to understand, learn, and apply knowledge to different tasks at the same level as a human.
Question 7 What is considered a key milestone in the evolution of Generative AI? A) The invention of the smartphone B) The release of GANs (Generative Adversarial Networks) C) The rise of social media platforms D) The discovery of quantum computing Correct Answer: B Explanation: GANs, introduced by Ian Goodfellow in 2014, revolutionized the field of Generative AI by enabling highly realistic data generation. Question 8 Which concept is central to many Generative AI applications? A) Hard-coded rules B) Manual data labeling only C) Learning patterns from existing data D) Avoidance of deep learning techniques Correct Answer: C Explanation: Generative AI relies on identifying patterns in existing data to generate new, similar data that was never explicitly seen before.
Question 9 Superintelligence refers to AI systems that: A) Excel in only one narrow domain B) Exceed human cognitive abilities across many areas C) Cannot adapt to changing data D) Are used solely in space exploration Correct Answer: B Explanation: Superintelligence is a hypothetical AI with intelligence surpassing that of the brightest human minds in virtually every field. Question 10 Which historical figure played a foundational role in the conceptual development of computing machines and AI? A) Steve Jobs B) Alan Turing C) Nikola Tesla D) Bill Gates Correct Answer: B Explanation: Alan Turing’s work on computation and the “Turing Test” laid the groundwork for how we conceptualize artificial intelligence.
Question 13 A key challenge in developing superintelligent systems is: A) Lack of computing resources entirely B) Ensuring these systems remain aligned with human values C) The impossibility of training any neural networks D) That no one researches superintelligence Correct Answer: B Explanation: One of the major concerns about superintelligent AI is controlling and aligning it with ethical and beneficial objectives for humanity. Question 14 The term “Generative AI” primarily implies AI systems that: A) Are solely text generators B) Are unable to learn from new data C) Generate new content rather than just analyzing existing content D) Focus only on numerical predictions Correct Answer: C Explanation: Generative AI involves creating fresh outputs—like synthesized images, text, or audio—rather than simply classifying or predicting from input data.
Question 15 Which of the following contributed heavily to the rapid growth of Generative AI in the 2010s? A) The decline of internet usage B) The development of advanced neural network architectures and the availability of large datasets C) A ban on GPU computing in research D) A significant reduction in computing power Correct Answer: B Explanation: Breakthroughs in deep neural networks and access to large-scale data and computational power fueled the rise of Generative AI. Question 16 What was a major driver for AI research in the 1950s and 1960s? A) Massive datasets collected from social media B) Military and government funding C) Blockchain technology D) The search engine revolution Correct Answer: B Explanation: Early AI research benefited significantly from military and government interest and funding, driving initial advancements in the field.
Question 19 What is a major benefit of understanding historical AI milestones? A) Avoiding data collection B) Identifying repeated pitfalls and guiding future research C) Preventing any new AI innovations D) Ensuring AI remains secretive Correct Answer: B Explanation: By studying past breakthroughs and failures, researchers can learn what strategies are effective and avoid repeating mistakes. Question 20 Why do we need ethical considerations in the development of Generative AI? A) To ensure it never improves B) To align AI outputs with human values and reduce harm C) To keep Generative AI behind a paywall D) To prevent AI from generating anything new Correct Answer: B Explanation: Ethical frameworks help guide responsible AI deployment, making sure Generative AI aligns with societal values and minimizes negative outcomes.
Question 21 Which statement best describes the evolution of Generative AI? A) It stagnated after its introduction B) It has rapidly advanced alongside improvements in deep learning C) It replaced all other AI techniques immediately D) It focused only on symbolic reasoning Correct Answer: B Explanation: Generative AI has significantly grown in parallel with breakthroughs in neural network architectures, computational power, and data availability. Question 22 What is one reason historical AI milestones are crucial for new AI researchers? A) They show that research is no longer needed B) They provide context, guiding innovation and preventing repeated failures C) They mandate the use of outdated algorithms D) They discourage new explorations in AI Correct Answer: B Explanation: Historical milestones educate new researchers on successful techniques, failed approaches, and help refine future research directions.
Question 25 Which type of learning involves finding hidden structures in unlabeled data? A) Supervised learning B) Unsupervised learning C) Reinforcement learning D) Transfer learning Correct Answer: B Explanation: Unsupervised learning explores unlabeled data to uncover underlying patterns or groupings without explicit output labels. Question 26 Which of the following is a key characteristic of reinforcement learning? A) Learning based on direct instructions only B) Learning by receiving rewards or penalties for actions in an environment C) Completely ignoring trial-and-error processes D) Removing any notion of agent-environment interaction Correct Answer: B Explanation: Reinforcement learning involves an agent interacting with an environment and learning through feedback in the form of rewards or penalties.
Question 27 Which technique is commonly used in deep learning to update model parameters? A) Manual reprogramming B) Gradient Descent C) Random data guessing D) Hard-coded feature extraction Correct Answer: B Explanation: Gradient Descent calculates the gradient of the loss function and updates the model’s parameters step by step to minimize errors. Question 28 Which algorithm is a classic example of a decision tree-based method? A) K-Means B) Random Forest C) Linear Regression D) Perceptron Correct Answer: B Explanation: Random Forest builds multiple decision trees and aggregates their results for improved predictive performance.
Question 31 What is backpropagation in the context of neural networks? A) A forward-only data pass B) A method to randomize weights C) A technique to compute gradients and propagate errors from output to input D) A technique to enlarge the dataset artificially Correct Answer: C Explanation: Backpropagation calculates how errors change with respect to each weight, adjusting them in the reverse direction to reduce the loss function. Question 32 Why are optimization algorithms crucial in machine learning? A) They ensure data is not used B) They randomly shuffle model parameters C) They efficiently find the best parameter values to minimize loss D) They remove the need for data preprocessing Correct Answer: C Explanation: Optimization algorithms like Gradient Descent are essential for efficiently converging on the best parameter set that reduces the model’s error.
Question 33 Which type of machine learning focuses on learning to predict continuous values, such as house prices? A) Classification B) Regression C) Clustering D) Reinforcement Correct Answer: B Explanation: Regression tasks aim to predict continuous-valued outputs based on input features. Question 34 What defines a Convolutional Neural Network (CNN)? A) Usage of recurrent connections for text generation B) A network that only processes tabular data C) Layers that learn spatial hierarchies of features through convolution operations D) A method that uses only one neuron for classification tasks Correct Answer: C Explanation: CNNs apply convolution filters that capture spatial relationships in data, particularly in images.
Which is a common challenge for large neural networks? A) They require almost no data B) They do not need optimization C) They can suffer from overfitting if not properly regularized D) They never require computational resources Correct Answer: C Explanation: Large neural networks can memorize training data rather than learning generalizable patterns, leading to overfitting. Question 38 Which is an advantage of using deep learning architectures over shallow models? A) They handle small datasets better B) They automatically learn hierarchical feature representations C) They rely solely on manual feature engineering D) They are always simpler to train Correct Answer: B Explanation: Deeper layers progressively learn more complex representations of the data, often reducing the need for manual feature engineering. Question 39
Which activation function can cause the “vanishing gradient” problem if used extensively in deep networks? A) ReLU B) Sigmoid C) Leaky ReLU D) ELU Correct Answer: B Explanation: Sigmoid activation can squash values into a small range, diminishing gradient magnitudes and causing the vanishing gradient issue in deep layers. Question 40 In the context of machine learning, “training” a model typically means: A) Manually entering all possible outputs B) Adjusting a model’s parameters based on data to improve its performance C) Writing code without data D) Setting the model’s parameters only once Correct Answer: B Explanation: During training, algorithms iteratively adjust parameters (like weights in a neural network) to better map inputs to desired outputs. Question 41
