A problem of coded bias: how tolerant is AI?доклад на конференции