Mimic Scripting

London, UK | 0 comments

Of all the things we covet – energy, power, influence, possessions – there are some that cannot be easily increased. One of those is attention, the act of “withdrawal from some things in order to deal effectively with others” (William James). Without augmentation, attention is a finite resource that many still have to husband carefully, using strategies and tricks like music, meditation, and filters.

Given the effort and cost involved in focusing attention, it’s all the more galling to have to ‘pay’ attention for easily-automated jobs or for social signalling. Think of the billions of hours wasted every year in the past by people attending to meetings and presentations with only one speaker allowed at a time, where use of other devices considered to be rude, or simply sitting at a desk pretending to work; one wonders how any work got done in the hierarchical corporate structures of the 20th and early 21st centuries.

By the twenties, a significant minority of meetings were taking place via videoconferencing due to outsourcing and fuel costs; naturally, people were still expected to pay attention, but workarounds were much easier. Increases in processor power meant that real-time video manipulation became affordable by the well-off, and the sophistication of expert systems and the size of the corpuses available to them led to perhaps the most powerful attention-saving device yet – the mimic script. AI historian Leo Kandel from University College London explains:

“Imagine that you’re sitting in your home, taking part in a videoconference with a dozen other people. You’re not expected to say anything important apart from perhaps answering a few questions at the end, but you’d rather not spend an hour just staring into a screen and doing nothing else when you have plenty of other things to do. So you turn on your personal mimic script.

“The mimic simulates your face and body, reacting according to the context of the meeting by, say, nodding along when everyone else is nodding. While this is going on, you’re free to get some real work done, have a sandwich, look after your kids, play a game, or whatever. And if you do get called upon, the mimic will immediately alert you while buying you some time with a ‘That’s a good question’, or some such.”

It doesn’t sound sophisticated, and it wasn’t. Early mimics were little more than statistical pattern recognisers coupled with actual hand-scripting and 3D engines, but they were good enough to be used by several hundred thousand within a year, and heavy investment from companies like Leverage and Apple led to mimics being able to carry out simple conversations, first via text and audio, and then over video.

Naturally, many people – particularly those from highly structured societies and organisations – found the use of mimic scripts to be lazy as well as extremely rude, and it was a common refrain to hear of workers being fired for supposedly ‘slacking off’ at work with mimics even though, technically speaking, the ‘work’ was still being done. But opportunities abounded for these lazy pioneers; skilled users could handle twenty sales videocalls at once, dipping in and out as necessary and tweaking their scripts for each situation. Tangentially, mimics also marked the beginning of the end for online poker.

If that was all mimic scripting was used for – freeing up attention for busy or easily distracted service workers – it would hardly merit a footnote. When people started buying and selling mimics, however, a whole new host of applications opened up.

Mimic scripting, though useful as a labour-saving device, was vastly more valuable as a kind of emotional prosthetic or tutor for the affectively impaired. For some, paying attention to something to something or someone isn’t enough – you need to understand and respond appropriately to the situation. A mimic script provided by an expert was invaluable to everyone from doctors with poor bedside manners to those with Aspergers. And wearables allowed mimic scripts to be used everywhere, unobtrusively, using face recognition to detect who you were talking to.

Estee Lem, an early user of mimic scripts, explains how important they were to him:

“When I was growing up, I was incredibly shy – I found it really difficult to talk to people, even friends and family. I just never knew what to say. After a very uncomfortable few weeks when a relative died and I said some stupid things, I decided to put a general purpose mimic script on my glasses. It didn’t solve everything, but knowing that I’d always have a line, something to say, when I met people, it made me so much more confident. I kept on using it for several years until Agents came out.”

Estee’s mimic was designed by Al-Qahirah, a collective of actors, data miners, and programmers, who created low-cost scripts for non-commercial purposes. For a fee, users could hire active assistance from the actors, who would finesse scripts and help provide better emotional judgement in real time; many actors had previous experience as puppeteers of Speekys.

If conservatives found self-mimic scripts troubling, they found borrowed-mimics downright unnatural. How could you tell whether they person you were talking to, the person you were in a relationship with, was mimicking a script – or even worse, mimicking another live person? Asking people to remove their glasses wasn’t practical, especially with active contacts starting to enter the market, and ‘mimic detecting software’ was notoriously inaccurate (not that it stopped people from using it, often with tragic results).

Today, these assumptions of unitary and stable identity seem quaint, but they were fiercely argued at the time; you don’t give up thousands of years of tradition and law on a whim. Questions of liability had to be thrashed out in courts around the world and new applications had to be found before mimics could fully flourish; for example, while most users initially hired instant expertise, the greater effect in the long run was the sharing of expertise and the construction of amalgamated mimics, and of course, the benefits given to the impaired.

Technologically speaking, mimics were a far cry from agents and true AIs, but their appearance to users was similar. The decisions and attitudes formed from mimics – mostly positive – would later shape the fate of the next technological revolution.