When Brett King and I challenged the SIBOS audience in London to try to imagine what financial services might look like in the future, we used a variety of science fiction works to illustrate different themes. It was a fun way to get people to think about what technolgy might do to payments, banking and money itself. I chose to focus on William Gibson, but of course I could have chosen any number of authors to think about different aspects of the future. Woking’s very own H.G. Wells, for example. Or George Orwell. Or Aldus Huxley. So, as the Financial Times just asked, what kind of future are we really hurtling towards?
I loved Isaac Asimov as a teenager and devoured his work. Not only Asimov, but Heinlein and many of the other “Golden Age” science fiction authors who painted a vision of mankind advancing into an intergalactic future. But now, having seen the movie Alien, having played the Paranoia RPG and having read Neuromancer, I have to say that I agree with the view that Dick’s vision of the future seems more accurate than Orwell or Wells or Huxley. A future in which you don’t know what is real or not. Well, not even a future. It’s now. We are already in Dick’s world, where machines can be mistaken for humans and humans for machines. A world filled with deranged billionaires, thinking machines and political chaos.
Central to this world is the inability to distinguish between what is “real” and what is not. Music, for example. Spotify has just announced that it will identify artists who appear to have been generated by AI.
The service will flag to listeners if an “artist’s identity may be AI-generated“ and does not represent a real person. I’m not exactly sure what this means, nor how Spotify will attempt to distinguish some AI-generated identities from all of the other AI-generated identites, but srely this is the wrong way round. What should actually happen is that all artists should be labelled as being potentially AI-generated until such time as they are able to provide an IS_A_PERSON credential.
What difference this will make to the public, I’m not sure. The fact is that people are remarkably poor at identifying AI-generated content. When people are shown a mix of genuine and AI-generated images, audio clips and videos, they perform only slightly better than random chance at distinguishing the real from the fake. While us older members of society might harbour the romantic notion that people might want to listen to a real artist, watch a real actor or talk to a real bank clerk, it is not at all clear that that is true. Perhaps, as Jamie Bartlett put it, everything is fake and nobody cares.
If you think this is a slightly over-the-top reaction, let me tell you that you are wrong.
Still think that “real” matters? Check out Chinese actor Fang Taozi. She shares breakfast videos. She posts workout clips. She documents her daily routine, interacts with fans and has just signed advertising deals worth up toa quarter of a million yuan for each of sponsored posts. She got 400,000 followers on a leading social media platform in China in only two months,. The short-form drama (“miniseries” made up of episodes that each last for 90 seconds or so) that she stars in has had 250 million views. The success of Fang Taozi has triggered intense debate about whether artificial intelligence has crossed a threshold from novelty to genuine competitive threat for human performers.
ang Taozi doesn’t exist, of course: she is wholly AI generated and probably represents the future of the entertainment industy. The dynamics of these Chinese miniseries are extraordinary. There were some 128,000 new microdramas released in the first quarter of this year alone and more than 95% of them are AI productions. I really do wonder whether the average British viewer, watching Eastenders on their mobile phone on the bus, would really be able to tell whether any of the actors are human beings or software constructs. And to Jamie’s point, would they care?
Anyway, does it matter as it’s just a bit of fun. Well, mabye. The OECD Policy Observatory labels these short-form Chinese dramas an AI harm, noting that addictive viewing behaviour by elderly users has negatively impacted their health and social lives while “severe economic disruption” has hit traditional actors and production workers, with a 75% drop in live-action filming and widespread job losses.
(The US leads the miniseries market outside China, already generating $1.3 billion in revenue, and Paramount+ has just announced that is preparing to add micro dramas to its app in the coming months, but Hollywood is still using human actors. Not for long, I’m sure.)
Which brings me back to the point. In Dick’s world, machines can be mistaken for humans and humans can be mistaken for machines. In Do Androids Dream of Electric Sheep?, his 1968 book that was the basis for the film Blade Runner, human bounty hunters come to resemble the soulless androids they hunt down for money. In the movie, as in many of Dick’s short stories, they are unsure whether they might be androids themselves.
You know where I’m going with this. We need a global strong digital identity infrastructure that works for things that exist, such as people and toasters, as well as things that don’t exist, such as companies and AI agents. I don’t think our politicians and regulators understand the magniture of the problem or the priority of a solution. Indeed, as noted in the FT, our leaders spending their time worrying about people who actually have a strategy about things, such as the Chinese Communist Party, and not enough time thinking about the threats that are already undermining us, ranging from predatory imaginary realities and billionaires reshaping people’s information environments to rogue AIs that are being released into the wild by agents of chaos.
We all understand that without know-your-customer, know-your-business and know-your-agent in place we can’t get the society that we want. But where to start?