Romantic Machinations: The best & worst part of our self to love
Breaking up but still living together
May 19th, 2023
Today I woke up and two of the automation scripts I’ve run against the Twitter API for the last 4 years have seemed to stop functioning. I’ve been anticipating this for months, wondering how it would roll out to low-level users like me. I didn’t expect to be able to do this forever, knowing the API changes in the 2010’s cut off web integrations and modules I had running on the previous version of Twitter’s API when they updated it that last time. Over the course of a week, I watched as most of the accounts went dark, returning an immediate 401 error for service to the Twitter API.
At times the only thing as meteoric as a hype cycle is the dynamism with which people emerge to begin shooting something down. Artificial Intelligence has had a unique and exemplary moment, but it’s not the first in the industry’s decades-long history and won’t be the last. However, it isn’t likely to continue making incredible leaps that wow average users until it fits in with domestic life. I’m reluctant to use the phrase “AI Winter” here. Though I don’t believe we’ll experience an ice age before the next incredible technical advances creep into our everyday lives in remarkable and helpful ways, the current level of attention is as warranted as it is unsustainable.
Now, speaking of automating our progress into a background function, let’s get creative. There are departments that wouldn’t mind having media on-demand. Just a click away, no art school.
The Living Word
So many folks working in creative industries, really anyone in marketing and even many sales positions, have spent time cutting their teeth on copywriting. It’s an unavoidable requirement of promotional work. While that might not be innately poetic, one does find their self struggling to rephrase words and terms that have been written or read after seeing them a few thousand times. (Edit note: Case in point, originally “struggling” was another word that I’ve used a lot and spent time mulling over this decision. You could say I struggled.)
Building on the point from earlier, two groups that have definitely identified the value of generative models are marketers and creatives. They’ve begun to incorporate the technologies as swiftly as they have iterated over the last few years. I have gone on quite extensively in the last six months about the breakthroughs in crafting visual elements as well as editorial brainstorming. The point here isn’t to revisit those ideas as much as note them as a milestone, or perhaps a high-water mark for this cycle. Video has come a long way as well but in many senses it is more a continuation of the image generation work we’ve seen from the premier entities in the space like Adobe, Midjourney, and Stable Diffusion.
This is really just the first stage of integrating the toolset, like dropping an extra effect on a guitarist’s loop pedal. A complete revolution in process and workflows is emerging as new development teams seek to offer more of the experiences people have come to expect from software that doesn’t feature AI components.
Marketing Technology Overhauls the Creative Department
Our desire to automate the mundane is one side of a coin that disconnects us from our local environment and the opportunities that enables. Such as their capacity to make the repetitive instead an awe-inspiring experience.
There has been a lot of hype regarding generative artificial intelligence but I think there has not been enough layman’s quality distillation of the heavy lifting machine learning as a brute force mechanism is still capable of in regards to remixing our existing repositories of media and to amend additional layers of context to the databases of meta information that exists mostly unseen on servers beyond the views afforded the designated frontends of a website.
One product making use of the data and doing additional intelligent processing is Augie, which I recently demoed. It is essentially a video editing skin with machine learning that connects images or video clips to keywords or sentiments in your script and has an AI reader to provide the voiceover. Their platform doesn’t seek to conjure synthetic notions out of the Web’s abyss but instead to help us act like video DJs, recompiling the works of others and finding new and surprising moments hidden within the raw materials after adding variation through the lens of our intentions.
AI allows a first draft iteration to be a more fully developed first conceptualization of an idea in a shorter amount of time than was ever possible. There’s a lot of value to working through ideas that you will ultimately reject, if artificial collaborators can refine that phase of the process by studying our previous choices or providing mockup alternatives then we humans as the creative engines driving the process can explore how these outcomes make us feel as a result of the options being incorporated into incubating versions of our work.
There are a few parties involved that don’t want to let automated intelligence be freely pervasive, as with current software practices, the thinking is probably to place these tools within the context of a subscription service that uses microtransactions for plug-ins and licensed content.
The Poetry We Asked the Machines to Write About Us
Unlike the unconditionality of a dog, machines can only emulate our extant expressions of adoration. Waiting at the door as a result of a machine learning optimization for appreciation.
June 5th, 2023
Since those first days when I wrote that a few of my scripts started returning 401 errors, more scripts have turned out the lights. Only a few accounts are still passing the commands through to the API and triggering activity on the network. Lots of developer accounts are reporting similar issues as the paid tiers of the Twitter API are rolled out to users, in turn unplugging automated activities from posting or engaging in other ways.
This action comes in the wake of other social networks reducing their API access, often with much less resistance. I believe this is because it was well before we really understood what it meant to be able to interface with social network data in a longitudinal way. My earliest experiments in no way reflect where I went with scripted activity in the following years. This discussion about the value of our creative output on the web and the underlying datamines is really just heating up as AI becomes ubiquitous. There is a caste system emerging around those that can afford to automate and those that don’t participate in that computational functionality. Positioning a moat around this processing is, in my opinion, much of the reason why a developmental halt was called.
The dog can hop up into bed where it knows it always has a spot, the neighbors need to call first before coming over and international travelers need to check in at customs. Our technology is becoming like an apartment we rent from companies that require a customs check-in.
Closing Transaction
The goal is not to push a message that seems anti-AI in any way, just to temper the heat around the conversation in regard to the lofty aspirations we imagine to be just around the corner. Certainly, the last 16 months have seen rapid and marked expansion to the capabilities, and by effect our expectations have followed along with those proportions. I am constantly impressed with the research coming out of Meta AI, their generative audio project called Voicebox debuted recently and it’s incredible. They intend to open source many of their models while also making them viable for commercial use, truly incredible contributions to the community. It’s not lost on me that this social network which exists on user-generated content is endorsing automated content to keep users spending minutes in their apps.
As soon as generative video becomes as intuitive as the progress being made with other media, I believe we’ll watch the hype fade into the background as this toolset becomes as commonplace as word processing and data management. Last week my remaining operational Twitter scripts went offline, it was sad. Like spirit friends lost somewhere inside the silicon. I have portraits of them in my phone but they can no longer stretch across Twitter’s portion of the Web, my personal arachnids.
As the improvements downgrade from milestones to incremental progress, the current euphoria about the great potential of AI will subside. The technology won’t go anywhere, it’ll just disappear into the background like the grinding hum of air conditioners and electricity transformers on a summer day with a light breeze. Always on but just out of view.