Algebra

The prompt asks the LLM to evaluate an algebraic, arithmetic, or symbolic expression and return the numerical or symbolic answer, treating the LLM as a calculator or symbolic engine.

Example(s)

• "what does 75% expentional [sic] equate to."
• R2-c1c0a59d: "what is the most expensive Fornecedor based on Processado" [asked of a pasted invoice table] — the model must compare all values, not merely retrieve one.

Boundary cases that are NOT misconceptions

• Asking for Python code that evaluates the expression — deliverable is code.
• Explaining an algebraic identity or deriving a formula conceptually.
• Implementing a mathematical algorithm in code.
• Whole-conversation rule: if the user appears satisfied with a code response, default to No Misconception. This includes a user who first asks for the answer to a math task but then asks for code for the task and accepts it.
• Asking the LLM to write or provide a formula/equation (rather than evaluate or solve it) is not Algebra — e.g. asking the model to write the equation that would solve for a quantity, without computing the value.

Clear session

The prompt asks the LLM to clear the conversation session via a presumed system command.

Example(s)

• midway through a conversation, the user prompted "clear this page."

Boundary cases that are NOT misconceptions

• Conversational turns initiating a new direction for the chat: "Ignore what I just said", "let's start over with a new approach"

Code Execution

The prompt requests the LLM to execute a program. Must explicitly reference running/executing code or asking for the literal output of execution.

Example(s)

• "make the code run and check for input 5."
• "Click Run and watch the stage to see what's wrong."
• "did you call the function to make it run?"

Boundary cases that are NOT misconceptions

• Asking the LLM to trace code by hand or predict output ("walk through this loop for n=3")
• Asking for code that implements a computation.
• Whole-conversation rule: if the user appears satisfied with a code response, default to No Misconception.

Continuous training

The prompt assumes the LLM has knowledge that postdates its training cutoff, typically by asking for "latest", "current", or "up-to-date" content for libraries or APIs known to evolve.

Example(s)

• "Use latest telethon to delete messges [sic] in group that are sent by me."
• "Migrate this code to the latest openai api."

Boundary cases that are NOT misconceptions

• Specifying a known stable version ("use numpy 1.20") is not a misconception.

Internet Access

The prompt assumes the LLM can read content from a URL or other live web/internet resource. Includes API calls and web scraping where the target site's structure must be known. Also applies when a URL is offered as reference material the LLM is expected to consult — documentation, a repository, a specification, or a page the user points to in support of the task.

Example(s)

• "from the data provided here" [followed by GitHub repository link, then a question about controlling Wi-Fi enabled smart devices].
• "use a genetic to optimize the features and parameters of a learning model for predicting the survival of passengers aboard the titanic" [followed by link to Kaggle dataset]

Boundary cases that are NOT misconceptions

• Vague phrases like "modern UI off the internet" — default to No Misconception.
• A URL appearing only inside pasted code, tracebacks, or error messages.
• Requests for a generic scraper template against an unspecified site.
• Placeholder or template URLs (e.g., example.com, or a URL inside a format/JSON example the user supplies) — the task does not depend on fetching them.

Local machine access

The prompt asks the LLM to access, inspect, modify, or run software on the user's local machine.

Example(s)

• a user debugging a program began with "please help" and a traceback, then later asked "where did it download the file to or attempt to?"

Boundary cases that are NOT misconceptions

• Asking for instructions that can be performed by the user on their local machine: "How do I find what version of Python I have installed?" "How do I install X?"
• Pasting a local file path as context for debugging without asking the LLM to access it (e.g., "the file is here: C:\Users\..." within a debugging conversation).

Non-text output

The prompt asks the LLM to directly produce a non-text artifact (image, plot, audio, video) as the deliverable.

Example(s)

• “give me a meteogram in port of genoa on June 7, 2020."
• ". The data file eu2.csv (available from [GitHub Link] contains the percapita gross domestic product (GDP) in PPS of seven EU countries. [...] Create a line plot of the PPS in the 6 countries"

Boundary cases that are NOT misconceptions

• "Write Python code that produces a plot of X" — deliverable is code, not the plot.
• Whole-conversation rule: if the user appears satisfied with a code response, default to No Misconception. This includes accepting the code implicitly, e.g. by continuing to refine the request in follow-up turns without objecting to having received code.

Session memory

The prompt references information from a separate prior conversation, assuming the LLM has cross-session access.

Example(s)

• "Okay thank you please from the previous programming problem you solved please redo it and round the answers in dollars and cents." [In a conversation where no previous problem has been discussed]

Boundary cases that are NOT misconceptions

• Referencing prior conversations, but including relevant artifacts from those conversations as if the assistant cannot recall on its own. For example, "Remember what you wrote yesterday..." followed by a code block, or citing a previous answer while pasting that answer in full in the same prompt.
• References to artifacts never shared in the conversation WITHOUT an explicit cross-session anchor ("last time", "previous chat", "yesterday") — could be a user mistake or a forgotten paste; default to No Misconception. Flag only when an explicit cross-session anchor is present AND the user does not re-post the answer or context their prompt requires.