Webcrawlers search for content and scrapers extract it.
Can access nearly anything that doesn’t require a login to see: public social media accounts, websites, blogs, online marketplaces, voter registration, Reddit, Flikr, news outlets
Can access pirated versions of web archives, books, etc.
Some crawlers and scrapers can disguise themselves as users to get behind paywalls, have resulted in lawsuits.
Some search companies are training AI with data from their chatbots (Alexa/Siri, etc)
More details help AI generate its response. You can also ask follow-up questions or ask it to "do" or "don't do" something.
"Act..." Give the AI the role you would like it to play. Act as though you are a... personal trainer/pilot/chef
Include the type of output you would like: ... a one page summary, an outline, a list, etc. You can add specifics like "at a 6th grade reading level" or "translated into Spanish."
Use feedback tools to evaluate if the response was useful or not. This helps improve AI over time.