liveBook · Manning

Verbosity

As explained in Starting Data Analytics with Generative AI and Python

The tendency of generated output to contain more words or repeated phrasing than necessary.

Verbosity is the tendency of generative AI to produce more words than necessary or repeat familiar phrasing. It becomes noticeable when prompts yield long, padded responses instead of concise answers. section 1.1

Recognizing padded phrasing

A verbose response may repeatedly use language such as “unwaveringly delve into vast landscapes of the rich tapestry of intricacies,” even when a simpler expression would suffice. section 1.1

Training-data patterns or biases may encourage longer responses or familiar structures. Verbosity can also consume limited response tokens, so inspect generated output rather than assuming its length signals quality. section 1.1

How Starting Data Analytics with Generative AI and Python covers it

How deep it goes
Identifies the behavior, gives a characteristic phrase, and attributes it to training patterns without developing a measurement method.
Examples
Uses an exaggerated phrase about delving into landscapes, tapestries, and intricacies to illustrate repetitive verbosity.
What you can do after
Evaluate generated responses for unnecessary length and repeated phrasing.
How it connects
Links verbosity to generative AI models, prompts, training data, biases, and token limits.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Starting Data Analytics with Generative AI and Python. The section links point to the parts of the book it draws on.