The Future of Content Creation: Humans or Machines?
My four year old son wants a dog. He wants to call him “Singo” after the dog from his favorite Swedish cartoon Alfons Åberg. My two year old daughter also wants a dog and wants to call her “dog”.
I learn a lot from my young children about the early workings of cognition. How they reason, how they calculate, how they produce language. It’s humbling, entertaining and gives me invaluable insight into how the human brain works. Children are like little versions of machine learning. Always watch, learn, imitate, move forward.
They anchor their worldview in the intention-based contexts and outcomes we give them, such as being good, kind, creative, and caring. They are young and have yet to endure the many rites of passage that will one day determine who they will become as fully formed human beings – enabling them to reason, calculate and produce language based on their own past experiences.
After all, that’s what makes us unique, isn’t it?
Humans vs. Machines: The War on the Media
Today, we humans are entrenched in some kind of phony war with AI content creation bots.
As a human being who works in communications, this should make me nervous. But the truth is, if machines can do a better job than us, then who are we to stop them? Shouldn’t we encourage a society where the best woman, man or machine gets the job?
In its latest Connected Smart Machines report, Ericsson tested eight media creation concepts among early adopters of consumer technologies in several major cities, to see which content we liked best. Some of the content was created by humans, some by machines.
The results caused celebrations on both sides. A surprisingly few in five consumers said they preferred human content creation over AI content creation in more than half of the concepts tested. However, the same number prefers AI content creation to humans. A dead end, for the moment. Regardless of your view, the trend points to a total machine takeover of the future content creation market, which means that the machine will inevitably be able to identify and satisfy demand for content better than us humans.
But what about the creativity of machines? As humans, the way we reason, calculate, and communicate reveals a rich mosaic of our own experiences and conceptualizations. It informs who we are and fuels our creativity. Will machines ever be able to match this creativity and continually push us towards new artistic genres? Will machines one day be able to think like us humans?
I create, therefore I am
In his 1949 Lister oration, the esteemed Professor Sir Geoffrey Jefferson became one of the first to question the creative prowess of these so-called mechanical menin which he stated:
“It is only when a machine can write a sonnet or compose a concerto because of the thoughts and emotions felt, and not by the chance dropping of symbols, that we could agree that the machine is equal to the brain – that’s that is, not just writing it but knowing that she wrote it No mechanism could experience (and not merely signal artificially, an easy ploy) the pleasure of her successes, the heartache when her valves fuse, be warmed by flattery, being made unhappy by his mistakes, being charmed by sex, being angry or depressed when he can’t get what he wants.
However, a year later, in his groundbreaking article Computing machines and intelligencethe father of artificial intelligence, Alan Turing, dismissed this argument as purely solipsistic:
“According to the most extreme form of this conception, the only way by which one could be sure that a machine thinks is to be the machine and to feel oneself thinking”. The truth is, machines don’t necessarily have to think like humans because they are, quite literally, wired differently. They just need to be able to imitate humans.
In the same article, Turing first offered his imitation game or Turing test as it is known today, which provided the means to measure the human nature of intelligent machines. The test, based on human-computer interaction through an intermediary, is still used today to help set the benchmark for today’s connected intelligent machines.
The Imitation Game: AI Content
Thus, consumers predict that mass media will be increasingly influenced by automation by 2030. The future, in fact, may be closer than we think.
Last year, OpenAI rolled out the beta version of its GPT-3 language generator, which many consider the biggest step forward in creating human-like AI content. He was even proved to be very effective in producing fast media forms such as poems, fan fiction, press releases, pop songs, rap songs, tech manuals and even this full article in The Guardian Last year.
As impressive as it is (and it is very impressive), the developers are already seeing that it does didn’t quite pass the Turing testindicating that we still have some technological distance to travel before we can truly mimic human intelligence.
And even the most basic AI language generators – even if they don’t stand up to a long Turing test – have proven to be a pretty good imitation on social media platforms. The rise of fake news on social media and the many existential questions it has raised show that AI content creation, like any other technology, can be used as a proxy for negative or malicious intent. This is why, as human beings, we have a responsibility to ensure that responsible AI is based on reliable and ethical AI principles.
Content and creativity: human after all?
But where do we stand when it comes to slower, more subjective and studied art forms?
Well, that’s good news for us. At the moment, consumers still prefer humans when it comes to music, with 65% saying they prefer humans as writers and performers of pop music. However, the Connected Smart Machines report also reveals that six in ten of us think artificial musicians will be able to outsell humans in the charts by 2030 – imagine something similar to the French hybrid Daft Punk human-machine.
We humans have been using computing to augment music creation for decades. Even Turing himself used his 1951 digital computer generate music. Bowie was also a early proponent of AI songwriting who said that “…[gave] a veritable kaleidoscope of meanings and topics and nouns and verbs colliding”. Today, there are many examples of musical artists teaming up with machines to create new genres of AI Music.
The report also reveals that consumers today view movies, like music, as a domain of human creativity, with six in ten consumers saying they would prefer human movie producers to their AI-enabled counterparts. However, most of these respondents are apparently unaware that AI is already being used in the movie industry to increase human decision making. Another perfect example of humans and machines living and working in harmony, right?
The future of content creation
So am I nervous? Content is as much about creativity as it is about automation, and as long as that remains the formula, I think humans will always have a prominent role in future content creation. This is because creativity is rarely a solitary pursuit, but rather a collaborative effort.
As Michael Björn, research program manager at Ericsson’s Consumer Lab and author of the Smart Connected Machines report, puts it: “The future of content creation is indeed collaborative, but I think creators who collaborate with the AI will have an advantage. One interesting area where this is already happening at a mass market level is in science fiction writing. Renowned Chinese SF author Chen Qiufan recently won a literary competition in Shanghai against competitors like Nobel laureate Mo Yan with the new “The State of Trance” which included AI-generated passages.”
So who knows, when our kids grow up, they just might have their favorite human-machine musical artist, or their favorite human-machine author. This could all be within the realm of the possible.
Read the Ericsson Connected Intelligent Machines report.
Read the Ericsson AI story.
Read previous messages from Michael Björn: