{"id":609,"date":"2021-11-24T19:36:55","date_gmt":"2021-11-24T17:36:55","guid":{"rendered":"https:\/\/artxai.uwas.aalto.fi\/?page_id=609"},"modified":"2021-11-25T13:23:18","modified_gmt":"2021-11-25T11:23:18","slug":"artificial-gossip-by-vilma-tiainen","status":"publish","type":"page","link":"https:\/\/artxai.uwas.aalto.fi\/artificial-gossip-by-vilma-tiainen\/","title":{"rendered":"ARTIFICIAL GOSSIP by Vilma Tiainen"},"content":{"rendered":"\n

If AI was asked what it thinks about you, what would it say? If different AIs had a conversation about you, what would the result be?  <\/p>\n\n\n\n

“Artificial gossip” showcases the communication between AIs which are having a “conversation” about me. It starts with one AI, the Google Ad personalisation AI<\/em>, interpreting my behaviour and summarizing it into short assumptions about me. These were available through my Google account. I wrote a simple Scala program to choose three sets of three assumptions randomly.  <\/p>\n\n\n\n

Each set of assumptions went through the following process: <\/p>\n\n\n\n

The assumptions in the set were first given as input to an AI text-to-image tool<\/em>, generating an image. Then, the image was given as input to Google\u2019s image labeling AI<\/em>, which produced a list of possible labels (words that explain what is in the image) for the image. Three of the labels were randomly chosen with another Scala program I wrote. These chosen labels were given as input to the same text-to-image tool as before, generating another image. <\/p>\n\n\n\n

This process created a back and forth between the two Google AIs and the text-to-image AI, which resembles a conversation. The three \u201dconversations\u201d generated with this process from the three initial sets of assumptions are what constitutes \u201cArtificial gossip\u201d. <\/p>\n\n\n\n

<\/div>\n\n\n\n

First conversation<\/strong> <\/p>\n\n\n\n

<\/div>\n\n\n\n

Google\u00a0ads\u00a0AI<\/strong>\u00a0
Construction & Maintenance, Marital Status: In a relationship, Shopping\u00a0<\/p>\n\n\n\n

<\/p>\n\n\n\n

Text-to-image AI<\/strong>\u00a0<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n
<\/div>\n\n\n\n


Google\u00a0labeling\u00a0AI\u00a0<\/strong>\u00a0
Water Transportation, Leisure, Truck\u00a0<\/p>\n\n\n\n

<\/p>\n\n\n\n

Text-to-image AI<\/strong>\u00a0<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n
<\/div>\n\n\n\n
<\/div>\n\n\n\n

Second conversation<\/strong>\u00a0<\/p>\n\n\n\n

<\/div>\n\n\n\n

Google\u00a0ads\u00a0AI<\/strong>\u00a0
Philosophy, Action & Platform Games, Outdoors\u00a0<\/p>\n\n\n\n

Text-to-image AI<\/strong> <\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/div>\n\n\n\n

Google labeling AI<\/strong> <\/p>\n\n\n\n

Leisure, Bench, Animation\u00a0<\/p>\n\n\n\n

Text-to-image AI<\/strong>\u00a0<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/div>\n\n\n\n

Third conversation<\/strong> <\/p>\n\n\n\n

<\/div>\n\n\n\n

Google ads AI<\/strong>\u00a0
World Music, Investing, Pet Food & Pet Care Supplies\u00a0<\/p>\n\n\n\n

Text-to-image AI<\/strong> <\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/div>\n\n\n\n

Google\u00a0labeling\u00a0AI<\/strong>\u00a0<\/p>\n\n\n\n

Puppy Love, Photography, Paw <\/p>\n\n\n\n

Text-to-image AI<\/strong> <\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

<\/p>\n\n\n\n

\u00a0<\/p>\n\n\n\n

<\/div>\n\n\n\n

Tools used<\/strong> <\/p>\n\n\n\n

Text-to-image tool: VQGAN + CLIP used with NightCafe creator (https:\/\/creator.nightcafe.studio<\/a>) and a Google Colab notebook (https:\/\/colab.research.google.com\/drive\/1ZAus_gn2RhTZWzOWUpPERNC0Q8OhZRTZ<\/a>)  <\/p>\n\n\n\n

Ad personalization assumptions: Google Ad Settings (https:\/\/adssettings.google.com<\/a>) <\/p>\n\n\n\n

Image labeling tool: Google Cloud Vision AI (https:\/\/cloud.google.com\/vision<\/a>) <\/p>\n\n\n\n

IDE used for Scala programs: IntelliJ IDEA <\/p>\n","protected":false},"excerpt":{"rendered":"

If AI was asked what it thinks about you, what would it say? If different AIs had a conversation about you, what would the result be?   “Artificial gossip” showcases the communication between AIs<\/p>\n

Continue reading <\/use> <\/svg>ARTIFICIAL GOSSIP by Vilma Tiainen<\/span><\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/pages\/609"}],"collection":[{"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/comments?post=609"}],"version-history":[{"count":14,"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/pages\/609\/revisions"}],"predecessor-version":[{"id":846,"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/pages\/609\/revisions\/846"}],"wp:attachment":[{"href":"https:\/\/artxai.uwas.aalto.fi\/wp-json\/wp\/v2\/media?parent=609"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}