All Roads Lead To Porn – A personal ramble about ChatGPT erotica

3 weeks ago 1

I’m both unsurprised and astonished that Sam Altman has announced ChatGPT will provide generative erotic services to its mature (over 18) users. Disappointing.

Just three months ago, I was visiting family in London with my girlfriend. While minding my business on the toilet and scrolling on X, formerly Twitter, I tried to click on some gory war video, but was met with an ID verification checkpoint to view it. Mid-toilet, I can’t grab my ID, and honestly, I only clicked the video out of habit. I am happy to skip another gore video, but I was intrigued by the small font text at the bottom of the verification step, which read something like: Let Grok verify you with a quick picture. Perfect, I thought. I’m mid-toilet and 25, so maybe I’ll get through Grok’s magic age guessing system. I didn’t get through, and Grok claimed I was under 18.

Growing up and living in NYC, I had never experienced this verification process or restriction on content. I didn’t mind it either; I could have viewed it if I wanted to. I have a legal license and am of age. But I wasn’t bothered. I looked into the purpose of this new verification step and what it restricted, and of course, mostly porn. How unsurprising!

I grew up with the internet. My programmer dad showed and explained to me early on how dangerous and fun the digital world can be, and he was correct. When I look back at my experience with the internet, it was definitely not just positive. I viewed Porn super early; I must have been nine years old. All it takes is a friend's older brother to expose the youngest in line, which always happens. Knowledge of the URL is easy to come by, and once you elevate from trying to search for kissing on YouTube to naked people on Google Images, PornHub stands around the corner waiting for your cursor to click. Access to porn is inevitable, so yes, I’m not surprised by Altmans’ decision to build ChatGPT erotica and “revolutionize” the porn industry with personalized and generated porn. The whole thing is unsurprising.

Free porn browsing websites are peak dual-use technology. The freemium model has allowed companies like MindGeek, a Canadian adult entertainment parent company that owns: Brazzers, Pornhub, RedTube, YouPorn, Tube8, Xtube, Digital Playground, Men.com, Mofos, Nutaku, Reality Kings, Sean Cody, TransAngels, Twistys, and WhyNotBi.com. A clear monopoly that uses oppressive algorithms to keep users watching porn for longer and sells harvested data through TrafficJunky. MindGeek makes millions by offering unique ad-targeting opportunities and selling user profile data to companies willing to target the adult entertainment industry. Now, tie that up with starting these data profiles on people like me at nine, and you have some high-value, organically created, hyper-personal user data. This data is some of the most valuable because of how honest it is.

Porn is lucrative! Websites like Pornhub may feel anonymous while using the service account-free, but there is a silent transaction between the user and the website: Deep and robust data collection/tracking. Pornhub will track user session times, pause locations, ad-viewing times, video preferences, etc. A data profile is made that ties personal sexual preferences with ad-targeting (Link). Now, take the average age of first-time porn consumption, being 9 and 13 years old (Link). Porn is undeniably easy to access with little to no age verification processes being implemented until recently, and is still hard to come by here in the US. ChatGPT is the best data collection service, but is also reasonably unprofitable. OpenAI operates ChatGPT at a loss but claims the target by 2029 is 125 billion dollars in revenue. If OpenAI’s operating costs increase, which they will as more users rely on ChatGPT, offering a paid subscription Porn Generation service seems like the most sensible thing to do.

With OpenAI’s new video generation model, Sora, users can generate a life-like video with sound attached. Since its release, my Instagram has been flooded with vulgar yet hilarious content of famous people doing silly things, saying awful things, or committing criminal acts. Although they make me laugh, when I heard that OpenAI was moving toward erotic content, my brain couldn’t help but imagine how video generation, fan fiction, and porn were all about to collide and explode in some kind of digital masochistic orgasm. Today, ChatGPT collects user data and uses it to train LLMs (large language models). How you respond and interact with ChatGPT is used to “improve” the LLM output and make the responses more human and tailored to your needs. This means, like porn, tons of user data and organically grown data profiles from hours spent being honest about your problems and needs. Slowly, ChatGPT has moved from a task-completion tool to a gray blob with less purpose and more human reliance (Link).

Last week, OpenAI announced the integration of product purchasing into the chatbot experience. This means users can now buy products through chat and be recommended through chat in the same session. I imagine OpenAI will quickly open up some kind of ad placement or marketplace page with “recommended” or “top” products sold on the ChatGPT marketplace. Of course, this list will not accurately represent what other ChatGPT users buy; instead, ChatGPT displays a table of probable products you will buy. When I shop on Amazon, tons of recommended products cover my screen based on intrusive data collection and ad profiling across my Google search and Amazon order history. Still, I never spend time talking to Amazon about an issue I may have with my roommate or an argument with my partner, followed by some random coding task I need to complete. Amazon is Amazon, I know what I’m getting when I surf products. With ChatGPT, it isn’t far-fetched to imagine the LLM trying to make me purchase items when trying to solve problems conversationally, and with OpenAI operating at a loss, it may make sense for the “human replacement” tool to end up being the world’s most unified, private, and intrusive marketplace.

Porn and adult content are the natural next step. If done “correctly,” OpenAI would be putting a stamp on internet monopolization, being the first to revolutionize the scope of dependence since Google. Why? Once you can generate your secrets, there will only be one website to solve everyday problems with a personal assistant that can use the context of your everyday issues and chat history to personalize your generated porn. How quickly does this data get sold? How quickly will OpenAI let others use the API for porn generation across other porn websites like those under the MindGeek umbrella? I’m very disappointed with ChatGPT and OpenAI. I’m also disappointed with humans for falling for the same trick over and over again. Altman started ChatGPT through scare tactics and fear-mongering, telling the world that AI was here and dangerous. If not controlled by Sam Altman and OpenAI, new AI technology could take over our lives. Now, three years later, reading the headlines about Sam Altman and the move towards erotica simply confirms that OpenAI and Sam Altman don’t, and never have had a human-first design philosophy.

What is Human-First Design
Simply put, Human-first design means building technology that is not dual-use. PornHub and ChatGPT are both great examples of dual-use technology because of their potential use cases outside of the users’ use of the website. For instance, PornHub provides tons of funny written data insights across many different topics, from Canadian election day voter data to traffic changes during a festival in Italy. PornHub can analyze user data to make real-world findings about non-related non-erotic subjects or events since users overlap (Link). ChatGPT is now flipping this relationship onto the user: create your own porn, create the trend, and then collect the data post-hoc. Data made this organically is hard to come by, but the format of conversation AI chatbots unlocks a treasure trove of organically made data. Harvesting, selling, and profiting from this data is where ChatGPT becomes dual-use. The same models that predict our answers help companies stop unions and fire the weakest link. Building human-first would mean avoiding these practices altogether. With tools as powerful as LLMs and video generation models, ChatGPT’s initial purpose and golden truth have continued to fade as use cases that sit far from utilitarian are projected to be more profitable.

How unsurprising.

Work Cited

https://fightthenewdrug.org/heres-what-porn-sites-do-with-all-the-info-they-gather-from-consumers/

https://www.newsweek.com/this-is-how-many-americans-are-addicted-to-porn-survey-10826171

https://www.pornhub.com/insights/

https://www.forcepoint.com/blog/insights/does-chatgpt-save-data

My Website

Discussion about this post

Read Entire Article