In mid 2020 I took a break from work. I was burnt out and exhausted, both physically and mentally, and —in the midst of Covid 19—I wanted a short break before starting on my new role as CTO of a Blockchain network analytics company. During those couple of months in the Summer of 2020, I was reading a lot about Machine Learning, now more commonly called AI. I was aware of transformers, but had never worked with them (my last major project in the field had ended about a year and a half prior and involved RNNs and LSTMs). Still, the power of the technology and the promise of GPTs were already apparent to me—and a short two and a half years later they rolls become apparent to everyone else too. I started writing “The Reviewer” as a futuristic dystopian short story informed by my thoughts after reading about transformers, GPT and their capabilities. I never finished it, as life happened and soon after I started working full time again, as it does, but I wrote enough to capture a slither of the fear, radical societal change and domination AI might bring in the coming years. Although unfinished I thought it made sense to publish it now, in its imperfect form. There’s no point waiting. Many of its predictions are already happening and it rings true more than it ever did when I first wrote it, almost four years ago. With much of its prophetic luster gone, I doubt I’ll ever sit down to finish it.
Still, I hope you enjoy it!
The Reviewer
Less than an hour remained before the show started. The whole family was excited. This was, after all, their first movie outing in a very long time. And this time it was special for everyone: they were going to watch the first movie wholly written, directed and produced by a machine. The first complex piece of art completely authored, produced and performed by an Artificial Intelligence. This wasn’t just another one of those CG-heavy Hollywood movies of yore. It was something else entirely: it was a story completely written by machines. Actors that were the sole product of a machine’s imagination. Places that never existed, anywhere on earth, but seemed so familiar, at least to the average human. It was, literally, snooping into the dreamworld of an entity the world had never experienced before. No one had seen this movie before; not even the producing studio, or the financiers that put together the millions of zollars required to gather the immense computing power required to ‘produce’ it.
Dad, a veteran movie critic in the local newspaper, was – paradoxically – enthused. He was one of the last living movie critics in the state. And at 57, he was way past his prime. Four, five full-length reviews a week was his best, while the money he got was barely enough for subsistence, let alone a comfortable life or, for that matter, providing for his family. For the past five years he wrote for a couple of publications with minimally overlapping audiences – a requirement, if he wanted to get around the unavoidable exclusivity restrictions his employers placed on him. After more than ten thousand movies reviewed, doing this full-time, for the better part of thirty years, it was a lot to take.
Even back in the 2000s, when Dad started working at the local newspaper, writing movie reviews as a junior critic, it was already arduous work for not exactly very rewarding pay. There was already so much information online and so much competition by amateurs and professionals alike and it was hard to differentiate and make a name for yourself. The Siskels and Eberts of the 2000s were hard to come by, and in less than a decade it became a non-starter, anyway, along with most entry-level writing gigs. But you still had a chance at making a living doing what you loved, and Dad truly loved movies and was a good critic. He watched every single one of them more than once, kept notes, digested their message, their æsthetics – the essence. Then he distanced himself from them, shielded his thoughts and wrote. Passionately, but fairly, soulfully, but rationally. His reviews were highly acclaimed across the state and, for a brief period, before machines took over his job, nationwide. His fifteen minutes of fame came a decade ago, after writing the ‘script’ for the award ceremony for a great director’s lifetime achievement award. Some, limited, recognition followed, including a collaboration with the Film Institute that lasted for the better part of five years, before that, too, succumbed to the incomperable prowess of the machine.
Twelve years later it was near impossible to make a living as a movie critic. Most of his colleagues had either retired or changed professions. Besides writing, dad had taken up extra work to make ends meet: assisting high-school kids with math and physics (he always had a knack for this) and maintaining some of the machines at a local launderette franchise. The launderette was fully automatic, from the till robot to the cleaning robots and, of course, the washing machines. Many of the washing machines were older models, however, from the late 2020s, and needed regular servicing. Dad knew how to mend them, was trustworthy and much cheaper and faster than the certified mechanic. He was also handy with SMT electronics, having worked as an apprentice at a local factory during his teens. Reflowing chips and soldering miniature capacitors was easy for him. And much more inexpensive than the board replacements the service centre mandated for, practically, every kind of fault possible.
But Dad’s passion was still the movies and he was increasingly distressed at his perceived inadequacy at reviewing them. It didn’t matter how good you were anymore, or how much time or effort you put at it. And not because there were so many movie critics around. The main reason was the rise of TheAuthor®, one of only three AI-based platforms that practically dominated the writing industry. TheAuthor was by far the most successful of the three. It was easy to use, affordable and extremely configurable. Say, you wanted a sophisticated, thorough, philosophical analysis of a dramatic film? Done. Complete with cultural references, subtle hints, but no spoilers, about the story and comparisons to previous works. How about an inspiring presentation of a high-tech adventure? No problem, technical jargon included that was sure to hit a nerve with the most relevant demographic. The kind of review that was statistically appealing to everyone: it would please and engage even the most hardcore techies, without alienating the general population. The text that would make your medium attractive and, given enough readers, the movie successful. If you were looking for something more commercial, that was easy too. TheAuthor was quick with uncritical, well-written presentations of new features, full of hype and enthusiasm; your ‘paper might even be able to get this one for free, if the movie studio chose to ‘sponsor’ it. There was increasingly little, if any, room for humans in the writing business, let alone being a professional movie critic. But it wasn’t just movie critics that were hit by the rise of TheAuthor. Amazon was full of books by TheAuthor and its competitors, newspapers were in some cases more than 80% automatically written by machines. National newspapers boasted that their content was ‘reviewed by humans’, but even they had passed the responsibility of actually reviewing the content to machines.
The developers of TheAuthor, the aptly named Authortronics Inc., had created other software services of this kind based on their original work: Product reviews and brochures, automatic summarizers, services for the government. Last year, for example, twenty four percent of the police precincts in the country used their products to produce their case and weekly reports automatically. You just fed video recordings from squad cars, officers, recordings and audio/video interviews and it automatically produced detailed, searchable reports for much of the less important, trivial paperwork. The police saved millions each year, not to mention considerable time for its officers, and Authortronics made billions as the undisputed leader in AI produced content in the world in the 2030s.
A spinoff of Authortronics, aptly named MovieGen Inc., founded by a team within Authortronics that loved the Cinema, was behind the software that produced this movie too. The result of fifteen years of progress in deep learning that had distilled centuries of human endeavour into a complex piece of software running on the company’s vast and constantly upgraded cluster of powerful computers.
Penelope stayed a bit longer in the shower and now they were running a bit late, given that they had to pick Jimmy up from his house, across the park, a few blocks behind Chesterfield Avenue before going to the movie theatre. Jimmy was a good friend of Dad’s from the old days. Let’s just say he was less excited about the film than they were.
He had recently lost his job as a middle manager at a software company. He used to run a team producing software for enterprises, one of those boring, gray pieces of table and graph-laden services that other companies use to make procurement, hiring, logistics and operations going smoothly. It was one of the good ones, founded in the early 2010s and rapidly expanded to fill a niche: an integrated solution for SMEs that combined more than a dozen ‘cloud’ services in one, easy-to-use product. It did well, not just in the States and Europe, but also in the rest of the world, mostly due to its modular and localizable design that made it adaptable to different languages, customs and workflows. Jimmy, in charge of a specific unit for Latin America, had travelled to Brazil several times in the past; he was a proud man, a good professional and completely dedicated to his work.
And he wasn’t an exception either. Most of the people there were skilled and devoted. They had spent decades building services and systems to serve their customers. Yet his company stood no chance against their latest challenger: Corporal Technology. In the short four years of its existence, Corporal had taken 35% of the market. Founded by two Kenyans and a Chinese, the new company was in stealth for most of its first year of existence, ignored the market and its competition and, instead, spent four years building an impressively solid foundation for a particular class of software products: self-developing Enterprise Operations software.
Their product, like the TheAuthor, also running on an impressive network of machines, was not only serving Corporal and its customers, keeping and processing their data, analyzing their behaviour and workflows, producing numerous, insightful reports automatically, but had the ability to self-repair, fix its flaws and improve its architecture and data management routines, continuously. Yet the company’s greatest achievement was to incorporate learning functionality, allowing the software to come up with enhancements to its feature-set. Sure there were limits to what it could do on its own, but there was no company of humans on earth that could compete with this system in its game. Bugs got fixed in hours, if not minutes, smaller features and enhancements got added in days, instead of weeks or months. All without a single engineer writing, auditing or even reading the code. There was literally no way to compete with them. Where Jimmy’s company had 200 people to support its customers, Corporal only had 24. And not one of those people actually worked on the product. They all worked on the ‘brains’ behind the product, the foundations that made all of it possible. Last year, when the government of Germany mandated the completion of a new form for importing a number of products into Germany, Corporal’s system automatically added the feature to the software, targetting German users only, by itself, following a ticket opened by a user with a link to the German Government’s web page. Ironically, both that web page containing the German Government requirements, and the changelog to the software introducing the new feature were written by TheAuthor.
So Jimmy, along with practically everyone else in his division, got the sack, and his company was sold, cheaply, to an Indian conglomerate. It was rumoured that they also had an AI platform in the works and wanted to get the client-base quickly, so as to get the upper-hand on Corporal. At 55, Jimmy was jobless and despondent. His severance package was relatively good, but his life was bleak and miserable. His wife had recently divorced him after twenty years of marriage because ‘he worked too hard’ and his children were in college, their relationship in tatters. The chances of him finding another job that paid half as much as the one he’d just lost were beyond slim and he was too young to retire – recently passed legislation meant he’d have to wait another 12 years at the earliest before he could do so anyway. His only obvious option was entering the hated Minimum Living Allowance along with the other 45% of the population. Being ‘fortunate’ enough to work on software, he hadn’t — until now — ever felt the pain of advancing automation and was always too proud and too cynical when it came to the ‘beggars’; a derogatory term many used for the 45% that lived off the MLA.
Needless to say, his cynicism did not dissipate, but turned into grumpiness and hostility toward anything AI. This made any interaction with him somewhat abrasive and the trip in Dad’s ageing self-driving car, frustrating for everyone on board.