AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  A process of manually programming computers with rules for every possible scenario

  Why it's wrong here

  Manually programming every possible scenario is the hallmark of traditional rule-based software or expert systems, where a programmer encodes deterministic if-then logic. Machine learning solves the inverse problem: rather than humans defining rules, the algorithm discovers the mapping from input to output by observing many examples. This is what enables ML systems to handle patterns that are too complex or variable for humans to explicitly enumerate, such as image recognition or natural language understanding, which is fundamentally different from explicit manual coding.

• ✓

  A subset of AI where algorithms learn from data to make predictions without explicit programming

  Why this is correct

  Machine learning is a subset of artificial intelligence in which algorithms iteratively adjust internal parameters by minimizing a loss function over training data, thereby learning patterns that generalize to new, unseen inputs. This is fundamentally statistical rather than rule-based: the algorithm infers the mapping from features to outputs automatically, without human-provided step-by-step instructions. Core paradigms include supervised learning, unsupervised learning, and reinforcement learning, each relying on data-driven optimization rather than explicit programming.

• ✗

  A method of creating robots that can perform physical tasks

  Why it's wrong here

  Robotics is the field of designing physical machines with actuation, sensors, and control systems to interact with the real world. While robots can leverage machine learning for perception or motion planning, machine learning itself is purely a software methodology for pattern recognition and prediction, requiring no physical embodiment. Equating the two confuses an application domain (robotics) with a computational learning technique that can run entirely on a server without any physical agent.

• ✗

  A type of computer network for processing large datasets

  Why it's wrong here

  Distributed computing architectures, such as clusters running MapReduce or Spark, are engineered to store and process large datasets across multiple nodes, addressing throughput and fault tolerance. Machine learning is instead an algorithmic approach that can run on a single laptop or a high-performance GPU, and its identity lies in the statistical optimization of models, not in any particular network topology or data-infrastructure design. While large datasets may be prerequisites for some ML tasks, the core definition remains about how models learn, not how data is transported.

Go deeper

Related to this question