Ad Targeting and Manipulation - Berkeley Law€¦ · Hooked: How to Build Habit-forming Products by Nir Eyal. New vs. improved/scaled up • Machine learning isn’t magic • Most
Post on 15-Aug-2020
2 Views
Preview:
Transcript
Ad Targeting and Manipulation
Michael Carl TschantzInternational Computer Science Institute
Web browsing Advertisements
Ad settings
Ad ecosystem
Inferences Edits
Web browsing Advertisements
Ad settings
Ad ecosystem
Set the gender bit to female or male
Browse websites related finding a new job
Significant difference ads on news website(p < 0.000006)
0%10%20%30%40%50%60%70%80%90%
100%
$200k+ Jobs - Execs Only Find Next $200k+ Job
FemaleMale
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Show to males
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Show to high earners
High earners are male
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Show to both Show to females
Input
Output
Female users
Other advertisers
WebsitesGoogle
Maleusers
The Barrett Group
Show to both
Clicking Not clicking
Accountability• Regardless of whether Google is responsible,
it should be held accountable as the one entity that can explain why it happened
Accountability vs. responsibility• Providing an explanation
vs. worthy of blame/praise
Persuading vs. manipulating• Commonly cited difference: Misleading• What about addicting?
Hooked: How to Build Habit-forming Products by Nir Eyal
New vs. improved/scaled up• Machine learning isn’t magic• Most of the features of online targeted
advertising already exists elsewhere
top related