We’re unready for the coming deluge of video, audio, photos and even text generated by machine learning to grab and hold our attention
Charles Arthur maps the evolution of AI-created content until now and ponders where it might be heading. This is a useful reflection, touching on the rise of algorithmically organised sites, as well as apps and frameworks such as MidJourney, GPT-3 and GAN. Thinking about this, he has a stab at what might be next.
• You could hook up GPT-3 to MidJourney and get it to try incantations to produce pictures, and feed the output to GANs tuned to pick output that humans will like
• Once that’s working, try doing the same with the text-to-video generator hooked up to GANs tuned to pick output that humans will like
In the end, Arthur explains that the future is already here and is progressively taking shape around us.
All the disparate bits above might look like, well, disparate parts, but they’re available now (and that’s without mentioning deepfakes). The trees are here, and the forest might be starting to take shape. Here’s an example: a 40-page comic book about monsters, free for download (PDF), by Steve Coulson, in which all the images are drawn by MidJourney. It’s very, very impressive.
Spotify Wrapped has become an annual tradition, marking the change of seasons the same way beloved cultural staples like Starbucks holiday cups or Mariah Carey mark the holidays. But as Spotify’s feature rose in popularity, so did a growing discourse about algorithms, the use of which has become standard procedure on social media, and which Wrapped relies on.
Reflecting upon Spotify’s Wrapped, the yearly review, Kelly Pau reminds us of the place of algorithms and artificial intelligence embedded within these choices:
An algorithm takes a set of inputs and generates an output, the same way a recipe turns ingredients into a cake. For Spotify to rely on algorithms means it uses data from its consumers to generate music discovery delivered through playlists. Open Spotify’s home page and you can find any number of curated playlists that source user data collected from the app, from “Top Songs in the USA,” which aggregates collective data, to “Discover Weekly,” which draws from personalized data. To create these playlists, Spotify tracks the music you listen to, organizes it into certain categories, measures tracks against other listeners, and uses that information to choose what music to show you.
These choices and recommendations often come with their own sets of biases and assumptions around gender and mood. They help mold a ‘templated self’ or what David Marshall describes as a dual strategic personadual strategic persona:
Through a particular study of online entertainment reviewing, this chapter explores the emergence of a new strategic persona in contemporary culture. It investigates the way that the production of entertainment-related commentary, reviews and critiques online is increasingly defined by a complex relationship and intersection with what is described as a dual strategic persona. Along with a public presentation of the self as reviewer across multiple platforms, the new online film reviewer is also negotiating how their identity and value are aggregated and structure into algorithms.
Although these curations are designed to share, I am more interested in using them as a point of reflection. I am always intrigued about what they do and do not say about my listening habits this year, this goes with the regular recommendations as well. I actually wonder if Spotify Wrapped reflects the place that music has served at times this year, a form of fast food, consumed as a means of escape, rather than something to stop and consider. For me, this has led to more pop at times. In addition to this, my statistics are corrupted in that I often play music for my children.
What remains more difficult to predict are the qualities that make you truly distinct: your thoughts and beliefs, your personal history, the unspoken nuances of the relationships that have made you who you are, and the unbounded expanse of moral and imaginative possibilities that constitutes your own mind. Attending to those aspects of yourself is the work of a lifetime—and far from boring.
Here are three (free) online sites powered by algorithms that I found and tinkered with. I am going to use the paragraph I just wrote as the intro I just wrote to this post as the text that I want each site to turn into music (See words above). Each of the sites will use the same exact text.
It is also interesting to consider how AI can produce manuscripts. I can imagine a younger me using this to generate compositions to cheat learning. I probably would have claimed them as my own, rather than being hampered by writing in a language not foreign to a budding rock guitarist.
If you want to have wilder, curiouser thoughts, you have to avoid the industrial monocropping of big-tech feeds. You want an intellectual forest, overgrown with mushrooms and towering weeds and a massive dead log where a family of raccoons has taken up residence.
I feel that the apex predator reintroduced as a part of this rewilding exercise is the question of time and productivity. We worry so much about demands and deadlines, that we fail to celebrate the things we have already done? Is the problem with doom scrolling actually the doom of the algorithmic nature of the feed, rather than the serendipity of dipping in? Or are the two forever intertwined? Is the answer ‘Twitter social distancing‘ or a reimagining of how we consume and create?
As Thompson himself attests, one answer is building up your own feeds. This is something that I have discussed here:
I like how Doug Belshaw frames the challenge as being in part about extending your serendpity surface. For Belshaw, the question is whether you curate your feeds or are instead curated:
read more widely and don’t settle for the “free.” algorithmically-curated, filter bubble being created for you by advertising-funded services with shareholders. We should be encouraging learners to do likewise. Doing so may take money, it may take time, it may be less convenient, but our information environments are important.
Beyond feeds, books and searches, I am also interested in sites like The Forest which add a touch of the unknown too.
However, at the end of the day, the missing piece in the rewilding exercise is people actually writing in a public square together to somehow celebrate the collective weirdness. I guess I still live in hope.
Aral’s post Hell site reminded me that, while I’ve talked about deactivating and reactivating my Twitter account several times, I haven’t mentioned ways in which I’ve found to battle the algorithmic timeline.
Doug Belshaw unpacks three strategies for gaining control over the Twitter black box:
My kid started Jr. High last week. He couldn't stop talking about how much he loved his History teacher. This afternoon we found him in tears, overcome by stress and self-doubt. His grade for his first short answer homework: 50/100. …It was graded by an @EdgenuityInc algorithm.
Artificial Intelligence and other advanced technologies are now being used to make decisions about everything from family law to sporting team selection. So, what works and what still needs refinement?
Also, they’re very small, very light and very agile – they clap as they flap their wings. Biologically-inspired drones are now a reality, but how and when will they be used?
To Neguine Rezaii, it’s natural that modern psychiatrists should want to use smartphones and other available technology. Discussions about ethics and privacy are important, she says, but so is an awareness that tech firms already harvest information on our behavior and use it—without our consent—for less noble purposes, such as deciding who will pay more for identical taxi rides or wait longer to be picked up.
“We live in a digital world. Things can always be abused,” she says. “Once an algorithm is out there, then people can take it and use it on others. There’s no way to prevent that. At least in the medical world we ask for consent.”
Maybe the inclusion of a personal devise changes the debate, however I am intrigued by the open declaration of data to a third-party entity. Although such solutions bring a certain sense of ease and efficiency, I imagine they also involve handing over a lot of personal information. I wonder what checks and balances have been put in place?
It’s difficult to know, in the typical chicken-and-egg conundrum, the extent to which Amazon is driving the public discussion on race, or our public debate is driving Amazon sales. Are the “Black Lists” pushing traffic on Amazon to particular books, and then those books pick up steam through the Amazon algorithm and get even more prominence? Or are loafing critics and readers cribbing from Amazon? At any rate, the online behemoth continues to hawk products by prioritizing them according to strong sales history and high conversion rates. The tyranny of the algorithm worsens our collective mental sloth where race is concerned. This mixture of culture, publishing, and code conflates traffic analytics with quality, and algorithmic recommendations with urgency.
Rich Benjamin discusses some of the problems and limitations to lists of books responding to political turmoil, particularly the impact of recommendation algorithms.
when people talk about whether algorithms are good or bad, they pretty much always mean decision-making algorithms – something that makes a decision that affects a human in some way. So for example long division is an algorithm, but it’s not really having any decision making effect on society. We’re talking more about things like putting things in a category, making an ordered list, finding links between things, and filtering stuff out. And they might be ‘rule-based’ expert systems, in that the creator programs in a set of rules that the system then executes, or more recently machine learning algorithms, where you train an algorithm on a dataset by reinforcing ‘good’ or ‘bad’ behaviour. Often with these we can’t always be sure how the algorithms has come to a conclusion.
So what the book is really focused on is the effect our increased use of decision-making algorithms like these is having on things like power, advertising, medicine, crime, justice, cars and transport, basically stuff that makes up the fabric of society, and where we’re starting to outsource these decisions to algorithms.
You often hear about artists under-appreciated in their time, who don’t find recognition until long after they’ve died.
Little known Japanese composer Hiroshi Yoshimura was one of those people.
Despite being a pioneer of the unique genre of kankyo ongaku – ambient music produced in Japan in the 1980s and 90s – most of his airplay came from the speakers of art galleries, museums and show homes.
He died in 2003, with most of his albums sitting as rare vinyls on the shelves of obscure record collectors.
That was, until a few years ago, when Hiroshi suddenly found millions of fans in the most unlikely place – YouTube.
Miyuki Jokiranta explores the way in which YouTube algorithms promote certain types of music to sustain our time and attention on the platform. This is something touched upon by the Rabbit Hole podcast.
At the moment it’s not the tech that’s holding people back from such decentralisation but rather two things. The first is the mental model of decentralisation. I think that’s easy to overcome, as back in 2007 people didn’t really ‘get’ Twitter, etc. The second one is much more difficult, and is around the dopamine hit you get from posting something on social media and becoming a minor celebrity. Although it’s possible to replicate this in decentralised environments, I’m not sure we’d necessarily want to?
Doug, I find the ‘I don’t get x’ an interesting discussion. Personally speaking, I thought I got Twitter five years ago, but now I am not so sure. Has Twitter changed? I guess. But what is more significant is that i have changed, along with my thinking about the web. I therefore wonder how long dopermine model will last until it possibly loses its shine? In part, it feels like this is something Cal Newport touched upon recently in regards to Facebook:
The thought that keeps capturing my attention, however, is that perhaps in making this short term move toward increased profit, Facebook set itself up for long term trouble.
When this platform shifted from connection to distraction it abdicated its greatest advantage: network effects. If Facebook’s main pitch is that it’s entertaining, it must then compete with everything else that’s entertaining.
I am not exactly sure of the moderation associated with decentralised networks, but I am more interested in streams that we are able to manage ourselves.
There is no real solution — the algorithmic genie is long gone from its bottle. But we can be aware, and make some decisions about how what information we share and how we are being manipulated by technology.