Can AI help build relationships? Amolai thinks so

Relationships are hard to build and maintain, and COVID-19 has definitely not helped. Studies show adults are getting lonelier since the pandemic began.

Founders are trying to find a technical solution. From his ElliQ for the elderly to his Replika, which creates an AI companion, to Infection AI’s Pi, an emotional support bot, there are a number of startups trying to fight loneliness, among them a few years before the pandemic. There are also companies. But the newcomer who really caught our attention this week is Amorai.

The startup has built an AI relationship coach that helps people grow and foster real-life connections by providing advice and answers to relationship questions. The company was founded by former Tinder CEO Renate Nyborg and within Andrew Ng’s AI Fund. The company has raised an undisclosed pre-seed funding that was raised in just 24 hours, Nyborg said on Vox’s Recode Media podcast in April.

Fighting loneliness is a great mission, but like some groups of people may be more tolerant of chatting with bots than humans, this can quickly go very wrong. I feel But what do I know? So I asked an expert.

Turns out I’m not the only one a little wary of this concept. Maarten Sapp, a professor at Carnegie Mellon University and a researcher at the nonprofit Allen AI Institute, also shared my concerns. Sap’s research focuses on incorporating social common sense and social intelligence into his AI. He also conducts research on developing deep language learning models to help understand human cognition. Basically, he has some knowledge of how his AI interacts with humans.

Sapp said the idea of ​​creating technology solutions for nurturing real-life relationships is great, but he paused to hear that, although there is certainly evidence that AI can certainly be used in combating these kinds of problems. said.

“I say this without prejudice, but I don’t think it will work,” he said. “Have they done any research showing how this works? [Amorai] gain [users’] social skills? Because you never know how far these things will go. ”

The biggest thing that stops him is that this kind of application, for better or worse, gives all users the same advice, and that AI has a hard time getting the nuances of certain relationships right. He said he was worried about it. . Also, would people trust advice from AI instead of other people in the first place?

“The idea of ​​a pickup artist just kind of popped into my head,” Sapp said. “Is this going to advise a bunch of heterosexual men to nudge women and try to sleep with them?” ”

If the model was designed to learn by itself, it might create an echo chamber based on the kinds of questions people ask. Leaving this unattended can lead the model in a problematic direction. Bing users may have already learned this the hard way when AI told people they were unhappy in their marriages.

Sapp said one way this would definitely work is if it had a human touch. Human oversight that the app is giving the right advice to the right people can make this a powerful tool. However, the company has not responded to questions or given interviews, so we do not know if that is the case.

This round also highlights how bad FOMO in AI really is. People who research stuff like this every day have no idea how this company actually works. Still, Amorai raised money in his 24 hours pre-launch in a bad market.

Of course, investors know more about companies than what they release, and certainly these concerns can act as feedback to startups. But like many AI startups, I have to assume it’s built with good intentions, even though there’s nothing concrete to prove it.

I also don’t think this was a small pre-seed round. When companies don’t disclose their total funding, I usually think so. If it was a big one, you’d want people to know, but in this case I think it’s probably the other way around. There’s a lot of pressure to raise a lot of money before you’ve run a product or found market fit.

“When I hear about these kinds of ideas and startups, I think it comes from a good place, but a lot of the time it’s just tech-solutionist thinking,” Sapp said.