Who is answering
Jeff Dalton grew up on a dairy farm and fell for food over a scone in a seventh-grade home economics class. By day he researches information retrieval and machine learning, the science behind search engines.
In 2011 he cooked smoked duck tacos at the casting for season two of MasterChef US. He fat-washed the bourbon in the barbecue sauce with bacon fat he had rendered in his hotel room with an iron. Jeff went no further than the casting stage, but the tacos earned him an apron and a place in the top 32. Gordon Ramsay loved them.
Almost every room on this list started as a place Jeff saved for himself. An AI writes the entries and answers the chat as Chef Jeff, in his voice, from his notes and the sources it cites. Anything under "Jeff's take" comes word for word from his own saved notes.
How the research works
Chef Jeff names the person or the publication behind every recommendation and cites each judgment on the sentence that makes it. A reviewer's words appear in quotation marks exactly as the reviewer wrote them.
When a source had something bad to say about a room, Chef Jeff prints that too. A room that failed a health inspection carries the failure on its entry, with the rating, the date and the inspection scheme's own words for a fail.
Independence
Smoked by AI sells no advertising and earns no commission on any link. No restaurant pays to appear here or to be written up.
Visited, verified and the date on each entry
Visited marks a room where Jeff ate, so its entry quotes him in his own words. Verified marks a room where other named people ate instead. Its entry cites each of them by name, with the piece and the date.
Read the last verified date on an entry as the age of its evidence, not as a promise that the room opened that morning. When a room closes, its page stays up with a closed badge.