If a "g" is voiceless, then it's the same as a "k". If we are going by IPA notation, the only distinction between the sound represented by "g" and the sound represented by "k" is voicing.
What taejo is saying is that native Korean speakers will hear [g] and [k] as the same (they are allophones in Korean), however the sound [kʰ] (aspirated voiceless velar stop) will sound distinct to them compared to [g] or [k].
Aspiration is a large cue for English speakers as well in distinguishing between voiced and voiceless consonant pairs ([pʰ] and [b], [tʰ] and [d], [kʰ] and [g]), hence why partial or even full devoicing can occur in word-initial voiced stops like the /g/ in "game" (as yongjik mentioned), or why English speakers have such trouble replicating voiced and voiceless pairs in languages like Spanish that do not rely on aspiration.
g is "by default" unvoiced, but it becomes voiced between other voiced sounds (vowels or nasals). So in gagu "가구" (furniture), the first g is unvoiced but the second is voiced.
What complicates matter is that, in English, word-first g (as in "game") can become partially unvoiced, so g in English "game" is actually fairly close to the first g of "gagu"!
Mapping ㄱ/ㅋ/ㄲ to g/k is imperfect, and the single-double consonants are especially hard for non-Koreans to even hear the difference between as we simply don't make any distinction, but Koreans do.
Furthermore, the pronunciation on many consonants differs based on letter position in the word, medial pronunciation is markedly different that initial or final.
e.g. ㄹ might be pronounced d/l/r based on position.
