Could Androids Dream Up Elaborate Stories?

Chris Leo
LDN - 11 May 2015
Chris Leo
Could Androids Dream Up Elaborate Stories?

Did you know an increasing amount of online news is being written by robots? It’s true — most of the major players are at it. For the time being our precocious algorithms are tasked with delivering dry, matter of fact content, like a football team’s past match history or the finer details of a quarterly financial report. But much like their Borg brethren, the algorithms are adapting, becoming better at understanding syntax and parsing sentence meaning.

The promise being we’ll end up with powerful and resonant stories, which came into being simply by way of a tried and tested formula (arguably a tactic Hollywood has employed for years).

As our imminent robot overlords defile our sacred love of storytelling, a tradition which predates writing, it begs the question: could an algorithm write an entire market research report and pass for a human?

Let’s start the MR-Turing test.

The case for: a robot could get away with doing an awful lot, convincingly

Automation has been on the tips of most tongues in the industry for a long time; NewMR even named it a buzzword for 2015.

It’s easy to concede a fair amount of processes could be automated and we’d all leave a little earlier, with less errors to wake up to the next day. This hasn’t gone unnoticed, and the industry seems to be undergoing a schism at the moment of ‘service-led’ vs. ‘platform-led’.

On this platform-led side of the continuum, we’re seeing ever more sophisticated tools capable of managing fieldwork, data tabulation and compiling reports, with little human intervention.

Whatever direction the industry takes in the next few years, it will be hard to imagine the process driven parts of our profession not succumbing to the rising tide of automation, much like how Directors regale new Grads with horrific tales of hand-drawn graphs, many of us will explain the intricacies of the table checking process and be met with similar disbelief.

But once the geeky robo-brains have mindlessly crafted their finished report, what next? Could they could get their grubby metallic fingers on the keyboard and hammer out a good story to explain it?

The case against: the uniquely human paradox of meaning

I’m going to argue they couldn’t and here’s why: meaning. Meaning means a lot to many people. If you’re teenager, it probably means existential angst and reading incomprehensible philosophical tomes. If you’re working in consumer insight, your meaning is going to be wrapped up in your job and your first question when you see a market research report is probably going to be: so what?

That report is going to be read by a silly little species with a limited attention and memory span. Even if every point within your report is salient, without a central argument to tie it all together, it risks being forgotten. That’s where the human brilliance of storytelling comes in.

Sebastian Faulks, of Birdsong fame, said recently that “Good storytelling should be in the service of good ideas”, and this creative faculty of pulling disparate information together into an impactful narrative is where us humans thrive.

Current scientific thinking suggests our brains are unparalleled in their ability to recognise patterns. Stay with me, I’m going somewhere. Our consciousness is looking more and more like an emergent property of the brain’s pattern analysis of external stimuli (data), as glimpsed through the senses.

Our brains evolved consciousness akin to a story about how the world operates, so that we understand it and survive. As our language developed we began to explain the world to others as we knew it ourselves: as stories.

A market research report may be a litany of stats and figures that a machine can populate, but giving context to those findings-telling the end client what it all means-that is something machines may never be able to do. Chiefly because, that machine would need to understand the bigger picture and that entails taking into account variables which may not be part of their current programing.

In short, they’d need to think creatively.

What’s our future? Will Skynet become sentient?

As Big Data grows into Huge Data, automation will likely become a necessity.

The current divide between service and platform-led offerings is becoming more pronounced, and further specialism between the people who ‘just give you the data’ and those who ‘tell you what the data means’ seems set to continue.

Yes, we (probably) have automated articles on whether Brentford thrashed Blackburn Rovers, but we’re still a long way off an algorithm that creates something that challenges our thinking, or inspires meaningful action and if you’ve paid a princely sum for a research report, that’s precisely what you want it to do.

Ultimately, if we looked our dutiful automaton square in the eyes and politely asked “what does any of this report mean?” It would probably sheepishly answer “42?”.

Stupid robots.