In the con­text of Ar­tifi­cial In­tel­li­gence or ma­chine learn­ing, a ‘con­cept’ is some­thing that iden­ti­fies thin­gies as be­ing in­side or out­side of the bound­ary. For ex­am­ple, a neu­ral net­work has learned the ‘pic­ture of a cat’ con­cept if it can re­li­ably dis­t­in­guish pic­tures of cats from pic­tures of non­cats. That is: A con­cept is a mem­ber­ship pred­i­cate.