In the past year the use of AI to answer verbatim questions has been on the rise. We have been paying strict attention to this issue seriously as have many in the industry (see Stanford article here ) 

The providers of online sample participants have for the most part been equally concerned as this undermines their whole business model. (if you can't trust the opinions are from humans, why on earth would you pay to purchase sample?). We've heard panels trying to mix education (don't use these tools) and purging panelists that repeatedly get caught using AI.

Here at Dufferin Research, in addition to implementing technological solutions like prohibiting copy & pasting within a survey, we are seeking to understand why the use has skyrocketed.

It isn't just as simple as cheating in many cases, often respondents can genuinely believe they are providing "better" answers when they are not knowledgeable enough, or are unsure how to answer. Here education seems the only solution, it's better to say  -- I don't know, or am unsure, or lack knowledge about the subject -- then to generate a "knowledgeable" response via AI.

It is in the industry's best interest to get a handle on this before we are forced to return to face-to-face interviewing as the only reliable way of ensuring we are getting human opinions.