AI startups are selling images of realistic computer-generated faces, letting clients like dating apps “increase diversity” in their ads without needing people
just input your desired ethnicity and emotion. Great story from @drewharwell https://www.washingtonpost.com/ ... https://twitter.com/...
Washington PostDrew Harwell
Context & Ripple Effects
This Washington Post report is an early marker in the commercialization of synthetic humans: rather than deepfakes of real people, startups are generating faces from scratch and selling them as ad inventory, with ethnicity and emotion selectable like product features. Months later, Synthesia was building corporate-friendly deepfake uses such as multilingual training videos, showing the same underlying capability being packaged for business buyers.
Dating apps and other advertiser clients can now cast 'diverse' campaigns on demand without booking models or photographers, cutting production cost and lead time to near zero.
Stock-photo agencies and modeling pipelines that supplied diverse ad imagery lose pricing power to generators that produce unlimited faces at marginal cost.
Second-order effects
Consent-based likeness businesses like Hour One become the counter-positioning play: if brands fear backlash over fabricated faces, paid, licensed humans are the defensible alternative.
Ad platforms and publishers face rising demand to verify whether campaign imagery depicts real people, feeding the market for detection services even as their accuracy claims go untested.
Third-order effects
The face economy splits into two tracks — compensated, consenting likenesses versus unattributable generated ones — and brand-safety pressure will decide which one advertising consolidates around.
'Diversity' in marketing becomes decoupled from any actual hiring or representation, creating a gap between advertised identity and organizational reality that regulators and advocacy groups can target.
The trend: Marketing is shifting from photographed humans to selectable synthetic ones, with paid-consent likeness models and detection services racing to catch up to the supply.
New: Are your ads too white? Is your dating app overloaded with men? Tech start-ups offer a solution: AI-generated people. 10,000 fake faces (as seen below) start at $100 a month https://www.washingtonpost.com/ ... https://twitter.com/...
Well one more avenue we dont need people any more for so adverts, tv and movies can all be done by computers, what are we going to do with ourselves... #FUTURE_AND_PAST #technews https://twitter.com/...
This is a brilliant idea, what a great business I hate myself for thinking of this https://www.washingtonpost.com/ ... IS_A_PERSON, people, I'm telling you, IS_A_PERSON will be the most valuable credential of all. Why aren't banks in this business already?
Dating apps are the new Craig's List! AI start-ups are selling images of computer-generated faces that look like the real thing, offering companies a chance to create imaginary models and “increase diversity” in their ads without needing human beings https://www.washingtonpost.co…
Dating apps need women. Advertisers need diversity. AI companies offer a solution: Fake people. @drewharwell does it again, unearthing important WTF stories on his #AI beat! https://www.washingtonpost.com/ ...
“The fact that such software now has a business model could also fuel a greater erosion of trust across an Internet already under assault by disinformation campaigns, ‘deepfake’ videos and other deceptive techniques,” writes @drewharwell in @washingtonpost https://www.washingtonp…
Important story/thread by @drewharwell on the emerging market for fake people (as a “solution” for corps who have diversity problems they do not want to address legitimately). https://twitter.com/...
Because it's a new year and I'm constantly making mistakes...here's a good rebuttal to this tweet about how synthetic data could actually help systems become fairer https://twitter.com/... https://twitter.com/...
Excellent and Important reporting @drewharwell on the growth of AI-generated images for marketing & business purposes. This concerning phenomenon will no doubt be leveraged for further fraud & deception online and eroding what limited trust exists in digital images today. https:/…
Using AI/ML to artificially increase BAME/PoC representation might sound like a good idea. But in reality... 👏🏻IT👏🏼 IS👏🏽VERY👏 🏾PROBLEMATIC👏🏿 One of the risks is to further reduce work opportunities for BAME/PoC models. Representation is great. Respect is even better. https://twit…
Advertising is the greatest application domain for fakery. Because they know that it is all fake they are trying to sell anyways. https://twitter.com/...
this is so embarrassing and gross. tech industry continues doing the most to avoid interacting with and paying poc fairly... are we surprised though https://twitter.com/...
stock photo companies are using AI to generate fake faces and provide “diverse models on-demand” — just input your desired ethnicity and emotion. Great story from @drewharwell https://www.washingtonpost.com/ ... https://twitter.com/...