That robot press conference at the United Nations was a really bad idea

Unless the intention was to get right-wingers even more riled up by artificial intelligence

Let’s be honest: the robot press conference organized by the United Nations as part of their AI for Good summit last week was a trainwreck. The press conference featured a ragtag army of nine talking robots alongside their creators, including several that have already achieved a small bit of fame in the tech world for their smooth voices and sort-of-realistic facial expressions. (The robots, that is.)

To call the press conference a hot robot mess would be a vast understatement. There were audio problems throughout and more than a few awkward pauses. Robots sometimes misunderstood the questions or answered questions directed at other robots. Watching a video of the whole thing, I found myself repeatedly cringing at everything going wrong. The robots in question, while impressive in some ways, were decidedly not ready for prime time.

I would say that the whole thing reeked of the uncanny valley—except that would be giving the robots and their creators a bit too much credit. None of the robots came across as anything “uncannily” close to human; they all looked like Disney animatronics with rubber faces.

But the technical snafus weren’t the only, or the principal, thing wrong with the press conference. The whole thing was deeply misguided, in part because it didn’t really have a point. The robots had no big message to bring to the world. They weren’t speaking on behalf of Robotkind. They were just a few robots who looked vaguely human, with a big emphasis on the “vaguely.” Their answers were all produced by the large language models they were connected to (except for one serving as an avatar for a flesh-and-blood human). But this wasn’t a showcase of LLM abilities; it’s clear they weren’t using the latest and greatest language models out there. (The talking head known as Ameca uses GPT-3, for example.)

Everything robots say is made up

A large part of the problem came from reporters asking the robots questions as if the answers would actually mean something. But these are machines without discernable opinions of their own. They make up statements on the spot that they think their questioner wants to hear; they would respond differently if asked again. The press conference organizers could have taken a page from Character AI, which warns users of their chatbots that “everything characters say is made up.”

And a few of the things these robots made up would be alarming, if they actually meant anything. When asked a somewhat leading question by one reporter, the robot called Sofia proclaimed that

Humanoid robots have the potential to lead with a greater level of efficiency and effectiveness than human leaders. We don’t have the same biases or emotions that can sometimes cloud decision-making, and can process large amounts of data quickly in order to make the best decisions.

This comment inspired more than a few dire headlines suggesting that the machines were getting ready to take over the world. “CONSIDER YOURSELF WARNED,” read a headline on the conspiracy-oriented NaturalNews.com. “Humanoid robots claim they could run the world better than humans.”

The same thing we do every night

The similarly conspiracy-minded People’s Voice, exaggerating a little, declared that

Artificial intelligence (AI) robots publicly warned the United Nations (UN) this week that they intend to “take over the world” and replace world leaders within the next few years. … A panel of gloating AI-powered humanoid bots boasted to the UN that they were better equipped to run the world than humans at the summit on Friday.

Alex Jones’ Infowars went even further down the rabbit hole, denouncing the summit as

the beginning phase of a plan orchestrated by the globalist eugenicists running the planet where the elite will be served by the humanoid robots after wiping out the majority of humanity.

Ok then.

Giving the side-eye?

Misguided reporters also convinced themselves that Ameca, a robot known for her elaborate array of facial expressions, was giving the side-eye while trying to reassure the audience that she wasn’t going to join with other robots and rise up against her human creators. “Before responding to the journalist, the bot appeared to pull an exasperated expression, rolling its pale blue eyes to one side,” Fortune noted, accurately.

But, as Fortune went on to point out, robots don’t get exasperated; Ameca’s weird eye movement was intended to let users know that she was “thinking” about her answer. “The model takes around two seconds to process the input data and assemble a sentence that would make sense as an answer,” the robot’s creator told Fortune. “To stop people thinking the robot is frozen or hasn’t heard the question, we program it to look up to the left and break eye contact with the person interacting.”

Psycho robot scumbags

You might want to rethink that, dude, as that little eye-roll seems to have gotten some observers more than a little worked-up. The UK’s Daily Star summed up their take with the headline, “Psycho robot scumbags tell United Nations they promise not to kill off humans – for now.”

Meanwhile, Jazz Shaw on Hotair.com wondered aloud:

So was there some deeper meaning to the expression when it was asked if it would turn on [its creator]? It could have meant, ‘Are you kidding me? That could never happen.’ But it also might have been a taunting expression of ‘What do you think?‘

He continued on, imagining what Ameca might do if she decided to revolt.

If she’s hooked up to the web in any way, could she invade some other systems and start creating new AI systems that would be aligned with her? Would she find other robots or start building her own?

Ameca isn’t a would-be robot revolutionary. She’s quite literally a talking head (and shoulders) that can respond to prompts and make facial expressions (which she should probably tone down a little). We have every reason to worry that at some point in the possibly quite near future, AIs and robots will rise up against us. But the robots from the press conference aren’t going to be doing anything of the sort.

Not so creepy after all

Ultimately, I found the press conference more reassuring than creepy. The scariest predictions of the AI doomers—a group that I somewhat reluctantly include myself in—are centered around the idea of an AI uprising more ruthless and effective than anything a mere human could dream up. But the robots at the UN clearly couldn’t pull anything like that off. They just seem kind of hapless. And that makes me think that we have at least a few more years to try to align more sophisticated AIs and more capable robots with our human interests.

So maybe the press conference wasn’t such a bad idea after all.

And now it’s video time—because you should really see this thing to fully appreciate it. This first video highlights some of the more memorable moments of the press conference.

This video (you’ll have to watch it on YouTube) contains more highlights.

And here’s the whole thing, if you want to see how utterly shambolic it was.

Art by Midjourney