Why does ChatGPT give different people different answers?
ChatGPT answers can differ from person to person because it does not work from one fixed script. OpenAI says it can use a person’s memory, an approximate location taken from their IP address, and a web search it may run when a question would benefit from current information. Two people asking the same question can get different answers, so one check from your own account shows what your account sees, not necessarily what a customer sees.
Memory
When Memory is enabled, OpenAI says ChatGPT “can remember relevant preferences and details from your chats and other available sources”. Depending on plan and region, those sources include past chats, saved memories, custom instructions, files in Library and content from connected apps, such as Gmail. OpenAI also says memory features and controls “can vary by plan, region, platform, and workspace settings”.
Memory can shape a web search too. OpenAI says ChatGPT “may use relevant details to formulate a more useful search query”, and gives location or dietary preferences as details that can refine a restaurant search.
Location
OpenAI says ChatGPT may use “an approximate location based on your IP address” to give relevant local results, and that sharing a device’s location is optional and off by default. Its example is a question about restaurants near the person, which ChatGPT rewrites to name the area the IP address suggests. A VPN or network location may change the location it infers.
OpenAI’s tip for better local results is to include your city, neighborhood or postal code in the question. A customer who types a town may get a differently aimed answer from one who does not.
Live search
OpenAI says ChatGPT “may search the web automatically when your question would benefit from current information”, so whether a search runs, and what it brings back, depends on the question. It adds that people who are not signed in can also use web search.
Google’s own AI features vary too. Google says AI Overviews and AI Mode “may use different models and techniques”, so what each shows will vary, and that AI Overviews are shown only when its systems judge that they add something to classic Search, so they often do not appear.
What this means for checking
The related answer on checking what ChatGPT says about your business sets out the monthly routine. One run proves little because your own account may carry memory and a location that a customer’s does not.
OpenAI says that before starting a temporary chat you can choose whether it uses existing memories, custom instructions and plugins, and that choosing Unpersonalized skips them. It also says temporary chats do not create or update memories. Ask the question the way a customer would, with the town in it, run it on more than one day, and ask someone else to run it on their own device. Treat the results as a pattern, not a verdict.
If a customer tells you what an assistant said, ask what they typed. The related answer on what to say when a customer repeats something wrong an AI told them covers that conversation, and the related answer on why ChatGPT recommends a competitor covers why a rival may be named instead of you.
Sources
- OpenAI Help Center, “Memory in ChatGPT”
- OpenAI Help Center, “Searching the web with ChatGPT”
- Google Search Central, “AI features and your website”
Related answers
- How do you check what ChatGPT says about your business?
- Why does ChatGPT recommend my competitor and not me?
- What should you say when a customer repeats something wrong an AI told them?
GEO: recommended in AI answers
When a customer asks ChatGPT, Perplexity or Google's AI who to call, the answer names a few businesses. GEO is the work of making yours one of them.