AI foundations
Build a precise vocabulary before comparing products, models, or research claims.
Steps
8 steps-
Artificial intelligence
Set the broad frame and identify what the term includes.
Artificial intelligence is an umbrella term for computational systems performing tasks associated with intelligent behavior. The field predates modern neural networks and has repeatedly changed its dominant methods and evaluation questions.
-
Machine learning
Narrow the frame to systems that learn from data.
Machine learning builds systems whose behavior is shaped by data and an optimization objective. Learning performance depends on data, representation, objective, optimization, and evaluation—not the algorithm name alone.
-
Deep learning
Place modern neural networks inside the larger machine-learning field.
Deep learning uses layered parameterized functions to learn representations at multiple levels of abstraction. Depth is an architectural property; useful performance still depends on training data, objectives, optimization, and compute.
-
Foundation model
Understand the reusable model layer beneath many products.
A foundation model is trained broadly and then adapted or prompted for many downstream uses. Broad reuse creates leverage but also propagates model-level limitations into multiple applications.
-
Large language model
Focus on language modeling without treating every AI system as an LLM.
A large language model estimates distributions over token sequences using a high-capacity neural network. Fluent continuation is not the same operation as consulting a fact database, even when outputs appear knowledgeable.
-
Transformer architecture
Meet the architecture behind many current language systems.
The Transformer replaces recurrent sequence processing with attention and position-wise feed-forward computation. Parallel training and flexible token interactions helped the architecture become a foundation for language and multimodal systems.
-
Self-attention
Inspect the mechanism that connects information across a sequence.
Self-attention computes content-dependent interactions among positions in the same sequence. Queries and keys determine weights while values carry the information combined at each position.
-
Model evaluation
Finish with the evidence needed to assess a model claim.
Model evaluation measures defined behavior on specified tasks, data, prompts, and scoring rules. A benchmark score is evidence for that setup, not a universal ranking across cost, safety, latency, and real workloads.