Natural language processing (NLP) is the field of computing that enables machines to manipulate human language: translating, interpreting, answering, summarizing, and generating text. It belongs to machine learning, although its history began with hand-written rules and logical operations. section 1.1
NLP has developed from early ambitions for machine translation, through rule-based programs such as ELIZA, to systems trained on large amounts of text. The Turing test made conversational imitation a prominent benchmark, but passing it does not establish flexible reasoning or genuine intelligence. section 1.1
Modern NLP supports several important uses:
- Large language models answer questions, summarize passages, generate text, write code, and support conversational agents. section 1.0 section 1.23
- Transformer models and attention help systems emphasize information that is most relevant to the current task. section 1.23
- Current large language models commonly learn from web-scale text with self-supervised learning and are then fine-tuned with reinforcement learning. section 1.23
- Useful outputs remain uncertain: models can encode bias, produce hallucinations, and consume substantial energy. section 1.0 section 1.23
From ELIZA to ChatGPT
ELIZA used simple pattern matching, open-ended questions, and generic replies such as “Please go on,” yet users sometimes treated it as understanding them. ChatGPT later demonstrated a broader generation-based approach by answering questions, explaining text, and producing poetry or code. section 1.1 section 1.0
Table 1.1 Types of machine learning
| Supervised learning |
Unsupervised learning |
Reinforcement learning |
|
| Description |
The model learns by mapping labeled inputs to known outputs. |
The model is trained without labels and without a specific reward. |
The model learns from its environment based on rewards and penalties. |
| Data |
Labeled data |
Unlabeled data |
No static dataset |
| Objective |
To predict the output of unseen inputs |
To discover underlying patterns in the data, such as clusters |
To determine the optimal strategy via trial and error |
Compare supervised, unsupervised, and reinforcement learning—the broader learning approaches connected to NLP systems. section 1.1
