(so bad)

Jan 3, 2023
∙ Paid

So now we know that fake news is bad, but Chat news is worse. Here’s some totally fake stuff, generated in realtime by writing it down in this Substack text field:

AI clouds can autosynthesize promocoins that appear to be about something, and then sell them in openair markets for time shares in Hawaii-lite vacation packages. By the time you return from your AI vacations, your house will be taken over by waiting families who watch text to stream news from nonexistent world crisis hotspots.
CostCo coins get you access to virtual AI glasses that project 3D realistic versions of 75 inch screens that you can literally walk out of the store without paying or even driving to the store. With no gas costs, you can call an Uber to deliver the glasses and some fast food along with some AI pets that don’t actually have to eat or utilize p-pads. As long as you agree to watch some programming to activate your behavior choices, the AI stream will feed you and register your votes in the inevitable midterms and runoffs. No need for messy fingers or going to the office; the office comes to you.
Next-gen AI glasses will take advantage of the proximity to the nose. AI comedians will benefit from auto-generated applause and heckler validation. AI tragedies will pipe auto-tears in at various stages, from moist remembrance trances to vibrant tears of rage. Wind puffs will simulate driving in a light breeze or splashing salty spray in a midnight walk along the moonlit beach. Noise cancelled AirPods will auto-select and tune out specific people in meetings to reflect a personalized and sanitized version of what you want to hear. For mandatory office attended meetings, an AI drone will project your presence without requiring any real action on your part. Hybrid services can let you attend several meetings in time-sliced fashion, or take phone calls while appearing to generate appropriate facial reactions in team meetings.
AI kibitzing channels let you join snarky comment sprees without letting on to the careless that you’re clocking them. This Twitterization of social attitude may be marketed to like-minded snarkers who can join from anywhere. Leaderboards emerge for laughter and tightening of the skin around the eyes to denote controlled anger and sarcasm storms. AI class clowns can record without others noticing and even autoedit promos and upload them to Substack TikTok clones. Trade behavior bursts as compensation for your favorite subscriptions. 
AI Boss will save you countless anxiety attacks, easing your worry about what the boss might think. AI generated bosses can be programmed to avoid boss-type ego storms that require managing up. Unleash customer bots that randomly predict the breaking point where there is no option other than churning away from AIVOD before your detectors are rendered inert by AI deflectors. And if in doubt, venture behind your Subscriber Only churnwall where everything is OK again.

It seems the problem here is a run-amok chatbot stirring things up with randomly factual harvests from Wikipedia. I didn’t like Wikipedia for a long time, doubting the ability of the faceless masses to compete directly with encyclopedias and the citation process. But slowly and surely the accumulated evidence of first-party documentation of things I did know about replaced my skepticism with begrudging respect. Learning the politics of the editorial process set me back for a while, as I grew more perceptive about the collective ax to grind of entry authors and editors. It was reminiscent of the formation process of civilian standards, where domain knowledge and vendor sports competed for leverage in the emerging documents. Standards washing was one part technique and another part art. Somehow social media emerged from the XML wars by adding a crucial element — identity. Not so much the name of the author but that of the citer. As information proliferated, who aligned with it became a useful sorting mechanism for the credibility and authority of the publisher.

This is where Twitter became a useful idiot, as a time snifter defending against the firehose.The social graph imparted by the follow and retweet/like signals provided such a valuable tool for harvesting value that the essence was not necessarily the information but the weather produced by the journey through the information. These influence storms were an easy way of navigating past trends and branding to a sense of the flow of interest mixed with the signal to noise of value over time expense. Facebook distorted this effect by making it clear the company’s imperative was to keep users within the moat of family, friend, and increasingly, a loss of trust in the after effects 0f the strategy. Separating the personality of the founder or acquirer from the question of aligning with a platform is fraught with danger, but easier to weigh based on the nature of the contract between service and contributor. As one example, testing the boundaries of engagement with a musician, actor, or comedian has relevance, but whether you experience the essential talent is not necessarily gauged by what you think about what you know of the person. He or she may be funny regardless of how they get from point a) to b). Sometimes the opacity of the inner person is a part of the joke. I find Steven Wright’s humor to be kind and gentle regardless of how good or bad a day he’s having or had. How the artist protects the process is not all that illuminating except as an element to be buttressed. Dylan eludes responsibility for following leaders, but the imperative speaks loudly to what he’s about at some layer worth knowing.

The holidays have once again surpassed the incredible irritation of getting everybody into the room. We’ve been filming conversations about the past, while the present performs at the demand of no-one but the players. Our granddaughter is a delight: I’ve always been stunned by the resident person even at the earliest of photographs. With Mila, I can see the future her from Day One.

Just read an article about post-dubbing to remove the F word from a film. AI was harnessed to create a map of visual patterns that conform to audio patterns of replaced expletives. According to the producer of the de-F’d project, the delta between the original r-rated version and the AI’d PG 13 version created the budget necessary to process the footage with the AI libraries. Using the project as a template for future applications of AI to creative works suggests a path for building out the technology. Can the same analysis be used to create new works from the start? The answer appears to be yes.

Let’s frame out this film. The AI interface asks a series of questions:

  • Is there a main character or several?

  • If more than one, what are the relationships between them?

  • Do we experience them at several ages?

  • What is the elapsed time encompassed by the film?

  • What is the point of view of the camera over the course of the film?

  • Does the audience learn things that the characters are not yet aware of?

  • How do the characters learn about such information?

  • Which of these strategies are used — flashback, exposition, inference via dialogue or clues as to location in the time frame of the story?

  • Does the film contain or communicate information about the Beatles?

  • If analyzed, could the film suggest the point of view of the filmmaker was based on the perspective that the most important Beatle was John; Paul; George; Ringo?

  • If John, would the focus of the film be about his relationship with Yoko; Brian Epstein; Elvis; his father; the Monkees?

  • If Paul, Brian Epstein’s death, the decision to release Yesterday as a Beatles track without any other Beatle present, the impact of the Sgt.Pepper escape from the touring Beatle path; his relationship with Linda; his decision to go public with the breakup of the group.

  • If George, why did he stay in the Beatles after he quit the Get Back project; was he the dominant Beatle because he had no part in the John/Paul songwriting dynamic; was he the key comedic owner of the group’s persona; in a three shot, would he remain separate from the others in terms of eye direction (i.e. not looking at the others nor at a photographer); was he avoiding the camera when he quit the group (see you round the clubs)?

  • If Ringo, did he know he was the straw that stirred the group?

  • Can u create a map of the composite Beatles’ sense of humor visually focusing on the components of the 4; George’s sneer, Paul’s producer ear, Ringo’s calm and outlet for his ego, John’s ability to survive his caustic impulse

The use of AI in the 50th anniversary Revolver made the real case for the surge of AI in the waning days of the Pandemic and the settling in of the economic crunch. You can imagine how this worked: each instrument and voice was mapped to its signature, then filtered for the overall blend of the instruments spread across the stereo picture. The success or conceit of the ChatGPT seems to leverage the huge volumes of data and the inferences drawn from the mix of facts and flavoring. I have no way to know this empirically because Peter Jackson’s engineers have kept most details private. But the Revolver mixes have a palpable sense of cohesion bootstrapped on a transparent screen on the details of vocals and instruments. It almost feels like an earphone-based mix with room tone added. By this I mean overtones mixed in through synthetic speakers that the musicians can hear with no real data.

Ringo’s drums in particular are startling in their creative energy and consistent invention. If you listen to the decay in the snare, you get both the pop of the strike and the clean afterimage of the effect on the overall feel. In order to create the necessary data, you have to model the complete “instrument” as both the driver and the concurrent interactor with the rhythm section of bass and guitars. Vocals blend in much the same way with a cinematic effect produced as a combination of background artwork and motion of the medium shot context. A little echo in the film track, and the layering of music for emotional spice. The fact that we know the inner data of the Beatles’ sound and humor so intuitively from constant replays produces a sense of what the real core is with the layering of AI emotion. As with everything the band touched with their producer George Martin, the alchemy is somehow enhanced by the intensity of the enormous details of the process. We hear what they heard when they stood and sang next to each other, the AI making subtle adjustments of what each heard from their close but distinct location. Just a guess but we’ll learn more from the next record< Rubber Soul, and even more if future passes are taken on the rest of the catalogue, including the anniversaries already released.

Another guess, but when I listened to the end credits White Album version of the Glass Onion film, it felt thin but intact. An AI enhanced version would have been just a little better. Once we get past the fad of synthetic press releases, we’ll see the remix artists combine with the technology and produce new ideas and explorations. My vote for Beatles starts with the I Want You (She’s So Heavy) gateway drug to the place the group left with.

Watch with a 7-day free trial

Subscribe to Gillmor Gang to watch this video and get 7 days of free access to the full post archives.