Once a new technology becomes popular enough — or only worthwhile — somebody will try to become sex technology somewhere. 3D porn, sexy VR and internet vibrators all had their sun time. Now that AI technology has started to feel more real than science-fiction fantasy, a few companies are trying, particularly in the form of AI-enabled sex toys, to integrate them into their clients' sexual lives.
The thing about an AI sex toy depends on who you're asking. Is it a sex toy, like the Smart Lipstick that is looking for funds on Kickstarter, that will talk with you?
Or is it like a blowjob machine that can learn how to give the last blowjob to the intelligence concentrate on the background?
Many AI experts do not really consider much of what is sold as erotic AI devices to be AI. Annalee Newitz, author of the novel Autonomous and co-host of the podcast, told me, "You do not need AI to respond to voicing commands." She explained that many of the currently available "interactive" toys do not actually take advantage of machine learning.
However, even if we haven't really reached the point of intelligent sex toys, the different products that have artificially smart themselves offer some idea of what consumers could want from a sex toy enhanced by IA.
How could this look? Kyle Machulis, MetaFetish creator, and Buttplug.io, told me he suspects that "a toy, that you're starting to use and that's what most people fantasize about when they think about an AI sex toy without having to really even think about it.
And in fact, one of the earliest efforts to upgrade a sex toy using AI was just a promise. HUM, a prototype of vibrator which appeared on IndieGoGo in late 2014, taught consumers a vision of white vibrators which would adjust their vibratory mode according to the movements of their bodies. As the first artificially smart vibrator, from The View to Cosmopolitan it was discussed throughout.
"You don't even need to press the buttons because of the sensors," the designers of HUM explained on their Indiegogo page. "HUM senses your touch and movement, creating vibrations that change your body 's harmony. Simply enable it and use it.
The media hype around his crowdfunding campaign showed a desire for the kind of efforteless erotic experience that the HUM team had promised to deliver. Although the product had not been released — and its website is now dead.
In recent years, other companies have come up with the idea that advanced programming could improve our masturbation experience. In October the Autoblow company released a white paper documenting the attempts to use AI to enhance its product , particularly by means of machine learning to teach sex toys a way to give a porn-quality blowjob, as part of its sexual playback toy that works like an automated masturbation sleeve.
Sloan hired a team to watch, annotate and gather 109 hours of blowjobs in porn scenes to train their analog to do what, in theory, feels like a blowjob of porn quality.
In a Skype call, Brian Sloan, the Autoblow inventor, told me that the company's recent project as a means of using machinery to make an automated blowjob more comparable with the real thing — to escape the idea of "modes" for sex toy and to move to one where sex toys offer "experiences." Long-term, he said, he considers that AI and machine learning can be even more beneficial to sex toy users.
Sloan said he was considering "a product [learning] how to stimulate your reaction to the product based on you."
Even though these products present an attractive future where it is easy to open a box and apply a product to your genitals, they also face obstacles that might hinder the achievement of that experience. One challenge: What are the creators of these products going to do to provide information on their IA, exactly how mental sex looks? What do they think?
The decision of the Autoblow AI team to train porn film blowjobs on one level makes sense: Easy data collection. There is no lack of porn, and there is no doubt that someone would presume to be able to educate an AI on sex through thousands (if not millions) of hours of hardcore content dedicated to the film.
However, even aside from the issue of how precise porn reflects sexuality, the data pornography is capable of providing significant limitations. A camera invisible or obscured by external viewer is some of the most important parts of sexual stimulation.
Watching porn (including porn for hundreds of hours) can not tell you what a blowjob giver does with your tongue, or what is inside a vagina to make you feel so good about sex. Things get even more complicated in an act such as cunnilingus, where porn artists often have to choose something that feels good for their partner and what looks good on their camera. (Machulis recently explained the problems with the use of porn to teach sex toys about sex incredibly in-depth.)
Lioness, a on the market available rabbit-style vibrator, is now equipped with force sensors, temperature sensors, accelerometers, and gyroscopes that map the orgasmic response pattern of a user — all data which, theoretically, can be used by an AI to reverse the orgasm. Lioness may seem a better way to develop this type of device with its reliance on biometrics with visual data, but there are also a number of problems with that approach. "I love the idea that arousal signals could take your device," Newitz told me.
"The problem is that the limited research available on [arousal signals] shows that women do not always experience physical signs of excitement as excitement. They will be greatly excited — there's going to be gorgle, grafting — but when they ask the women, 'Are you excited? 'It's answer, 'No.'"
Indeed, several researchers have discovered that arousal by women (and sometimes men) is often inappropriate, which means that your brain and body do not always agree whether or not you are activated.
You may want to anticipate sex as your body does not show any physical excitement, or your brain may have no interest in sex whilst your genitals become swelling, lubricating or otherwise excited. Some research has even shown that many women are physically reacting to images of bonobos that have sex, without a sense of sex. This poses an important issue for an intelligent vibrator: how can we guarantee that, because someone has shown a picture of bonobo sex, what an intelligent vibrator registries as a user and is truly converted into a stimulation pattern is that rather than just a noise?
In the end, the very problem that they try to solve can hamper intelligent sex toys. The AI sex toy hype history indicates that we want a sex toy that we know what we like instantly, without effort, because we are either unwilling or unable to find it for ourselves. But if we don't make the hard work of figuring out exactly what turns us on and gets us off, we can't really teach sex toys how to get us to the highest peaks.
