Intelligence and Corporatocracy II: Trash Future

When SCSP put out that “blog post” the other day, I hadn’t planned on this synergistic point between AI and defense infrastructure becoming a theme. Yet — here I am writing about it for the second day in a row. This fixation precipitated from yesterday’s evening events after I put up the first portion. I went on a short walk, and three things of note took place. First, I listened to a podcast that I don’t normally. Second, the police were called on me. Third, I met my step goal for the day. Huzzah for the final point, let’s address the first two.

The title of the podcast is “trashfuture.” The commentators have a tendency towards left politics, are generally more-intelligent-than-average, and talk about topics that interest me. I saw that the most recent episode queued from them was regarding AI and the defense industrial sector. Fascinated by the ever-more-speedy march towards the terminator movies, I selected the podcast and began listening.

Setting forth from Jon’s house, I walked outside of the neighborhood’s gate and button-hooked behind the now-abandoned CVS. This CVS borders Westheimer Road, one of the busier (and shadier) thoroughfares that leads to and from downtown Houston. The CVS has been stripped a number of times for all of the available copper, rare metals wiring, cables, and the like by the local community of the unhoused. Most recently (this week) some less-than-legitimate actors pulled up in a white, un-badged Chevy pickup truck and took some corrugated metal sheeting from inside the delivery bay. I don’t think it’d easy to fence something like that, and if it’s the most-valuable-loot-available: I find it highly likely the CVS is fully picked-over. There are short waist-height fences (drawn in yellow) bounding the area, as well as some overconstructed 8–10' high fences (drawn in black) bounding the commercial area bordering Westheimer. I’ve attached a map below showing the route I walked (in blue).

Map of the evening’s events. My path is in blue, short/incomplete/passable fences are in yellow, large and impassible barriers such as walls or solid fences are in black.

I navigate unhindered past borders in the yellow fences, and I don’t bisect the route of any of the taller fences. I’m making these choices in order to avoid interacting with the local behind-CVS residents as well as minimize my exposure to Westheimer. Best decision I can make? Don’t go on a walk outside the fenced-in area where Jon and I live — but what kind of life is that‽ Next-best decision? Walk behind the businesses and in the direction of a bordering neighborhood. This area from CVS to the now-defunct Chuy’s is open, unmarked, and unbounded by any natural or plain artificial barriers.

As the podcast gets going, I’m sucked into the narrative they’re presenting. Apparently ChatGPT’s owning firm (OpenAI) has reevaluated their moral pause in relation to working with the defense industrial sector in the name of “preventing veteran suicide” (something we should all find plainly dubious). I’d like to address veteran suicide properly (we will in the third and final installment of this series)— we need to get this discussion about AI out of the way first. AI’s danger in the defense industry doesn’t singularly lie in it’s attachment to weaponry. In terms of Large-Language-Model integration with defense/intelligence, one need look no further than the information domain of the battlespace. Simply winning the informational fight is the highest priority for the middle ground between the Intelligence Community (IC) and the defense industry. These applications are a beyond-nuclear option simply because of the global addiction to and use of social media. The ideological commons of the globe now exists on the internet — an open-source domain of information that people willfully volunteer as publicly available. Using AI to process this “large dataset” of information is the only way that either of these communities can afford to do this, as we discussed in the first post of this series. The Air Force already has a fantastic tool they’ve aptly named “Dark Saber” (source). In the logo they’ve placed on this landing page, they’ve included a red light saber.

I’m a big Star Wars head from back-in-the day, and this piqued my interest. Please forgive my diversion. I also have experience in the military, so I know that this patch was made by some O-3 or O-4 with little-to-no talent for both graphic design or military leadership and a career that will likely peak with these design choices. The blade is lit and red, which is a fascinating choice depending on whose light saber it is. Heraldry matters. In this case, because it’s the organization affirming it’s own moralistic motifs. Look at the image itself — we also see binary code, a half globe (showing the Saudi Peninsula and the adjoining northern third of Africa), and a rocket shooting into space. All clear motifs of the type of “innovation” and global-focus scope that the Air Force is targeting. Why is the tool named as a weapons system? Why must it be evil? Why couldn’t they use a name like ALICE? It’s also worth noting (as supporting evidence) that the Air Force is predominantly staffed by nerds who know which light saber hilt they’re putting in the graphic. I, myself, remember sitting on the floor with my favorite member of the Air Force’s research-industrial complex (Jason) — and watching Star Wars Episode III for the first time. This in the same living room where Patton had been playing to an audience with pseudo-religious fervor.

There are four different nerd cabals that I interlope between, and I sent query to each of them regarding this. The first responded to me with general dubiousness, as I know them primarily through our shared affinity for cottage core. The second, my longest-held DnD group, also ultimately gave me few resources. The third and fourth groups — both centered on different eras in history (I prefer to keep my dabblings with Ancient Egypt and 19th Century separate from each other, although those Briish archaeologists are pesky) — both fruitless. If someone is willing to correct me, I’m content calling this either a creation of the artist or* a poor representation of the hilt belonging to Qui-Gon Jinn (played wondrously by Liam Neesons). In the first case: the creator of this image felt to make the most centrally representative icon of the tool it alleges to depict an icon of evil, malice, and individual selfishness. In the second case, the artist has chosen to corrupt a symbol of moral justness and fairness — in the tone of Qui-Gon’s character. Either choice that has been made is a knowing one.

The defunct Chuy’s has the right security setup to deter the same homeless population that exists behind CVS. Like the security barrier between me and access to “Dark Saber,” the automated response tools that Chuy’s has licensed/subcontracted from HPD are a hard and clear barrier. I cut corners with my walking path through the lot, and I thought I had greater than 50m between myself and the device. I did not, and thus the device activated. It lit up with blue flashing lights, and audibly stated in a clear and masculine tone “this area is off-limits! Authorities have been notified and are responding!” I already dress in clothes that can be described as both “layered” and “nondescript,” so I walked jauntily up the roadway pictured and waited. My impetus whenever I’m in a stressful/critical/dangerous situation is to look at my watch and make sure I’m keeping track of time.

Westheimer Road is not a safe place, and police are aware of that. Response time for a single SUV cruiser with sirens was inside of a minute. They were on-site for less than 5 minutes and left. They could have only been there for me, they didn’t find me, and they were satisfied. The “theft prevention systems” had cameras attached and a shadowy top-down angle of my face is probably on a (presumably) poorly-maintained storage system somewhere. However, I’m almost certain that the same footage and/or my face couldn’t have made it into the police cruiser at that time. AI shortens that information lifecycle. I have unique high-contrast features on my face (moles) that make my likeness excellent at training facial recognition software, as well as being highly vulnerable to targeting using it. My driver’s license has a clean-shaven photo of me with shorter hair, and my eyes haven’t changed since I was born (let alone 2016). Let me suppose (in the same way that the SCSP did) about a HPD Detective’s workflow integrating AI. They could implement a relatively simple program that scrubs the footage for any face with a high level of confidence (already in existence and use), and then have the product of that process be used as a query input for the driver’s license database.

That leads to an arrest by the police, and is measured as a successful outcome for all involved (hypothetically). However, we do not live in a binary or ordered world. We cannot make any existence ordered under the powers that humanity has in it’s reins today. Even Kardashev would say that ultimately humans will not ascend past his great filter. The algorithms that I modeled in the last paragraph can function as an “AI” in the same way that LLMs do. They measure (with the dataset they’re trained on) a given query against a large dataset that they have access to through computational deduction and give you an output that meets the per-arranged criteria with an acceptable level of confidence. That’s it.

I return home, dry off, and go to sleep. I’m fine because we don’t live in that world…right? I’ve been kept up late since because of Ring doorbell cameras and their history of gleefully surrendering footage to the police, because actors like the SCSP are more concerned with funding than the moralistic impact their actions have on America’s position in the current hegemony, and because of Dark Saber.

Like I mentioned yesterday — AI simply solves the man-hours problem too well for people to ignore it. All of the cognitive grunt work that had to previously be managed by entire humans can simply be completed with a prompt and some good-ole computational power. Let’s not forget that the Air Force — just like every other branch of the US military — is going through an enlistment crisis. Not a hiring crisis, mind you. The only thing AI is going to be implemented for is the establishment of another successive managerial tranche, and the DoD is no exception. In simple terms — officers don’t have enough enlisted personnel to complete their given missions, and then supplement those shortcomings with AI-augmented personnel that have “increased productivity.” These personnel then ultimately become — due to the reliance on ever-increasing mission efficiency and productivity creep-the taskmasters of more and more LLMs and task-oriented AI. Sure, the idea of a drone aircraft with lethal munitions that only has to be ordered — and won’t question the moral or ethical foundation of those orders — it’s horrifying. But not as horrifying as accelerating mission outcomes past the watchful eye of a citizen-enabled regulator.