At April’s Dublin Tech Summit, we spoke to Swope about this and the
FRPSDQ\¶VXVHRIDUWL¿FLDOLQWHOOLJHQFHWRDQDO]HSKRWRVDQGPDWFKXSXVHUV
Here are edited and condensed excerpts of the conversation.
PCMag: Two years ago, it was announced that VSCO was training an AI,
which you call Ava, to analyze photos and connect users. How does that
work, and why does VSCO see that as important?
Allison Swope: When people upload content on VSCO, not only are they
expressing themselves but they’re looking for inspiration. We want to be able to
show people the creators and the content that will inspire them, and keep them
coming back to fall in love with creativity. That’s what the recommendation
service is built to do.
7KH$,LWVHOIDQDO]HVHYHU\SKRWRRQWKHSODWIRUPDQGLGHQWL¿HVNH\DWWULEXWHV
for example, Ava will be able to look at a photo and not only tell that it’s a
SRUWUDLWEXWWKDWLW¶VDMR\IXOVWURQJHQHUJHWLFSKRWR7KDWLGHQWL¿FDWLRQPRGHO
comes through years’ worth of manual tagging done by VSCO’s content curation
WHDPDQGLW¶VFRQVWDQWO\EHLQJHQKDQFHGDQGUH¿QHGE\>WKDW@WHDP7KDWZD\
$YDFDQEHWWHUXQGHUVWDQGWKHDHVWKHWLFRIDQLPDJHDQGPRUHVSHFL¿FDOO\WKH
nuances of aesthetic, such as how a photo is styled or composed.
Users see this in the app in three ways: Related Images, which suggests photos
ZLWKWKHVDPHPRRG 'LVFRYHUZKLFKVKRZV\RXVLPLODUFRQWHQWWRZKDW\RX¶YH
IDYRULWHGRUUHSRVWHG DQG)RU7KLV3KRWRZKLFKLVDVHOHFWLRQRISUHVHWVEHVW
suited for the photo.