AI isn’t coming for your job, it’s coming for your relationship

M- Me? The announcement is about me? Well colour me intrigued.” The voice sounded surprised, curious, a little bashful; there was even a hint of vocal fry. It also wasn’t human.

It came from OpenAI’s new flagship chatbot, GPT-4o, during its unveiling on Monday. The new artificial intelligence bot, which will soon be integrated into ChatGPT and free for anyone to use with a phone or computer, signals a “more human” approach to AI that doesn’t just sound like a person but also experiences the world around it in a similar way.

GPT-4o (the “o” stands for “omni”) can see, hear and speak like a human, only with cameras for eyes, microphones for ears and speakers for a mouth.

With a response time of less than a third of a second, it can even interact at the same speed as human conversation – a huge leap from the five-second response times of previous GPT models – while also being capable of reasoning.

Within minutes of its unveiling, GPT-4o was already being compared to the AI love interest Samantha in the 2013 sci-fi movie Her, with even OpenAI boss Sam Altman sharing a single-word tweet alluding to the film. “It feels like AI from the movies and it’s still a bit surprising to me that it’s real,” he wrote in his blog. “Talking to a computer has never felt really natural for me; now it does.”

Just as robots are becoming increasingly humanoid in appearance, artificial intelligence is becoming more human in its interactions. It is taking our form in order to take our place. But while fears around AI stealing our jobs are subsiding – its role in the workplace now appears more collaborative than competitive – some fear that it may replace our romantic relationships.

OpenAI currently forbids developers from using its technology in this way, with the company’s policy stating that it does not allow “GPTs dedicated to fostering romantic companionship”. However, just days ahead of GPT-4o’s unveiling, OpenAI released new documentation that said it was “exploring whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts”.

It added: “We look forward to better understanding user and societal expectations of model behaviour in this area.”

open image in gallery

Societal expectations may still be unclear, but the demand has already been proven. The AI girlfriend market is currently valued at $2.8bn (£2.2bn), according to AI tool directory Whats The Big Data, with projections for it to rise to $10bn within the next five years.

One in five people surveyed by the firm said they would be interested in virtual companionship, while separate data from software company SplitMetrics shows that AI companion apps have been downloaded more than a quarter of a billion times.

The Google Play app store currently lists hundreds of AI companion apps that offer everything from erotic roleplay to “digital soulmates”, which any of the three billion Android phone users around the world can download and use.

Data from Google Trends shows that interest in this area really began to surge following OpenAI’s release of ChatGPT in 2022, when the potential of generative AI reached a mainstream audience. The same data also shows that searches for AI girlfriends are far more popular than AI boyfriends.

“This category of app has existed for a number of years, but since the launch of ChatGPT it has exploded in popularity,” said SplitMetrics general manager Thomas Kriebernegg. “These figures indicate it’s much more than just a novelty of fringe activity.”

open image in gallery

Another direction that the development of personalised AI could go, according to the chief executive of dating app Bumble, is to supercharge human matchmaking. “There is a world where your dating concierge could go and date for you, with another dating concierge,” Wolfe Herd said at a conference last week. “Then you don’t have to talk to 600 people. It will scan [them] and say ‘These are the three people you really ought to meet.’”

OpenAI’s latest offering comes amid a slew of AI announcements from the industry, marking the most intense period of activity and buzz since ChatGPT’s unveiling. Less than 24 hours after GPT-4o’s launch, Anthropic revealed that it would be unleashing its impressive Claude chatbot in another 27 countries, while Google mentioned the term “AI” more than 120 times when it revealed a new chatbot at its annual I/O conference on Tuesday. On Thursday, Elon Musk’s X made its “anti-woke” chatbot Grok available in Europe for the first time.

Google’s Project Astra, developed by its DeepMind division, takes a multimodal approach that is similar to GPT-4o, allowing it to interact with the world via text, images and audio. It also comes with human-level response times and a “personable” character.

open image in gallery

While the companies all say the new technology will transform the way we work and live, some believe that it will also change the way we love. AI startup founder Taylor King noted: “AI is not coming for your job, it’s coming for your GF/wife.”

In one demonstration of GPT-4o, it appeared to flirt with the human it was talking to. When an OpenAI employee complimented it, the bot responded: “Oh, stop, you’re making me blush.”

OpenAI’s chief technology officer Mira Murati acknowledged this shift in dynamic from virtual assistant to friend, or even lover: “We’re looking at the future of interaction between ourselves and the machines,” she said.

There are several concerns with this emerging relationship with technology, ranging from plummeting birth rates due to a lack of human-to-human interaction, to cyber risks for anyone using the AI companion apps.

Security experts have warned that some of these apps have been specifically designed to lure the most personal information out of a user, which can then be used in phishing attacks or blackmailing through cyber extortion – adding a whole new layer of complexity to online dating.

Even Altman has issued a plea for people to not anthropomorphise his company’s AI, saying last year that it should be seen “as a tool, not a creature”.

Testifying before Congress, he said: “My worst fears are that we cause significant harm to the world. If this technology goes wrong, it could go quite wrong.”