What you need to know about AskISI in CEIC.
Is AskISI generative AI?
AskISI uses a large language model (LLM) to interpret your question and work out which data answers it — not to generate the answer itself. Everything returned is existing CEIC data retrieved from the database. The model doesn't produce values, series or analysis.
Can it invent numbers or hallucinate?
Not in the data. AskISI doesn't generate values and can't invent a series — every result is a real CEIC series you can open, inspect and source-check as you do today. The numbers always come from CEIC, never from the LLM.
Does it replace the existing search and data tree?
No. It sits alongside them, and nothing is being taken away. If AskISI doesn't return what you need, existing search and navigation work exactly as they do today.
Will it only show me data I'm entitled to?
AskISI searches the full CEIC database to assemble the best answer, which may include series outside your subscription. Unsubscribed series appear greyed out — the title is visible, but the data can't be opened or downloaded.
Is our data used to train the model, or shared with anyone?
No. User interactions and CEIC content are not used to retrain or fine-tune the underlying model, and there's no external data sharing beyond the secure model API. Data in transit is encrypted, and access is role-based and logged for audit.
Can AskISI be turned off for my organization?
Yes. Reach out to your account manager and it can be disabled for your organization, or for named users.