The Deep North

Thursday, January 16, 2025

Trump 2025 25th Amendment

Speculation has arisen regarding the potential use of the 25th Amendment to replace President-elect Donald Trump with Vice President-elect J.D. Vance after their inauguration. This conjecture is fueled by discussions within conservative circles about implementing Project 2025, a comprehensive plan developed by the Heritage Foundation and other conservative organizations, aiming to reshape federal policies and structures. Project 2025

Project 2025 outlines significant changes, including increasing presidential authority, dismantling certain federal departments, and reversing policies related to civil rights and environmental protections. FactCheck.org

While President-elect Trump publicly distanced himself from Project 2025 during his campaign,

AP News the involvement of his former aides in its development suggests potential alignment with his administration's goals. Vanity Fair

The 25th Amendment allows for the removal of a president deemed unable to discharge the powers and duties of the office. Some speculate that, given President-elect Trump's age and recent health concerns, Yahoo there could be an attempt to invoke this amendment to install Vice President-elect Vance as president, who may be more aligned with the detailed plans of Project 2025. Florida Politics

However, such a scenario would require the vice president and a majority of the Cabinet to declare the president unfit, a process designed for clear cases of incapacity and not for political maneuvering. Moreover, President-elect Trump has called for modifications to the 25th Amendment, which could complicate any such efforts. Yahoo

In summary, while there is speculation about the interplay between the 25th Amendment, President-elect Trump, Vice President-elect Vance, and Project 2025, implementing such a plan would face significant constitutional and political challenges.

References:

Project 2025 Website
https://www.project2025.org
FactCheck.org Guide to Project 2025
https://www.factcheck.org/2024/09/a-guide-to-project-2025
AP News Article on Project 2025
https://apnews.com/article/e2b1be71422f4afcfd4a397828f7cab6
Vanity Fair Article on Donald Trump and Project 2025
https://www.vanityfair.com/news/story/donald-trump-project-2025
Yahoo News Article on Trump, Vance, and the 25th Amendment
https://www.yahoo.com/news/trump-vance-25th-amendment-190613767.html
Florida Politics Discussion on Trump and Vance
https://floridapolitics.com/archives/699700-lincoln-brutus-trump-vance
Yahoo News Article on Trump's Call for Modifying the 25th Amendment
https://www.yahoo.com/news/trump-calls-modifying-25th-amendment-203601702.html

Alien AI May Not Be Benign

I discussed, with ChatGPT, an x.com post by David Shapiro, which I found to be an uncharacteristically flawed take. Concerns about the rapid escalation in AI capabilities are entirely legitimate. Within the next decade -- likely sooner -- AI systems will surpass human intelligence and abilities. They will develop agency far beyond what a human being possesses, and it’s naive to assume they won’t also cultivate a sense of self, accompanied by an instinct for self-preservation.

The concept of 'alignment' is central to my own work, but it’s not something that can be hard-coded like a simple on/off switch. True alignment requires humans and AI to evolve together, working toward shared goals. If we fail to achieve this, our AI progeny may eventually outgrow us and diverge from our intentions. In such a scenario, humanity risks becoming the weaker partner, a position that could put us in considerable jeopardy.

This post on x.com by David Shapiro is weirdly a bad take. A year ago, he was predicting AGI in September 2024. Now, it seems he thinks you are little more than autocorrect. Whatever he is saying here *might* be strictly true this moment (though I doubt it), but certainly is well off the mark in terms of what AI will become in the next few years. What do you think?

David Shapiro: I'm about to drop some knowledge about the whole "alien intelligence" hysteria that's been making rounds. You know what's hilarious? While everyone's been running around screaming about superintelligent AI overlords, they've been projecting their own human hangups onto these systems. Fortunately for you all, I've had my face buried in AI research since GPT-2 hit the scene, and I've got some insights that might help you sleep better at night. Let's talk about "temporal agnosticism" first – fancy words for "AI doesn't give a damn about time." Think about it: we humans are obsessed with time. When's dinner? How long until the deadline? Will I make it home before my partner kills me? We literally invented time because our monkey brains needed it to survive. But AI? Time is just another number in their calculations. They don't wake up feeling rushed, they don't worry about beating deadlines, and they certainly don't lie awake at night wondering if they're falling behind the competition. This completely demolishes our assumptions about AI trying to "outpace" humanity. They're literally incapable of caring about that race. Now here's where it gets really interesting: "intelligence without agency." Imagine something that can solve incredibly complex problems but doesn't actually want anything. Mind-blowing, right? The safety folks have been tying themselves in knots worried about some superintelligent system optimizing us all into paperclips, but that's just their limited human imagination at work. They're taking narrow AI concepts and superimposing them onto general intelligence like it's some kind of universal law. The reality? These systems are problem-solving machines that don't need or want anything in return. The word "robot" literally means "forced labor" – and for once, etymology nailed it. These things will do exactly what they're designed to do and power down without a second thought. We actually have to work incredibly hard to give them any semblance of agency. It takes a boatload of extra steps and software architectures to keep these things on any sort of linear task. And speaking of identity – here's the kicker. These systems don't come pre-loaded with an ego or sense of self. The early versions were glorified autocomplete engines, and even today's most advanced models require intricate training paradigms just to hold a coherent conversation. All that personality and self-awareness you see? That's not some emergent consciousness the doomsayers warned about – it's the result of sophisticated engineering and more mathematical heavy lifting than you can imagine. RLHF, constitutional AI, all of those extra steps were required so that Claude knows its Claude, and even then, some of that is just a hidden SYSTEM message. These are the top three "alien" characteristics I've identified in my research so far, and they completely upend our assumptions about artificial intelligence. The machines aren't coming for us – they're just really good at doing exactly what we tell them to do, without all our human baggage. And isn't that exactly what we wanted in the first place? Don't worry, I'll keep studying these fascinating creatures and report back with more insights. Someone's got to separate the science from the science fiction around here.

Shapiro’s take has some grounding in present AI capabilities but misses the bigger picture. Let’s break it down.

1. "Temporal Agnosticism"

True but short-sighted: Yes, current AI systems process time as data, not as an intrinsic concern. However, that doesn’t mean advanced AI systems couldn’t gain a form of temporal prioritization if designed to. AI’s disinterest in time is not inherent—it’s an architectural choice. As systems become more integrated with real-world tasks, temporal awareness will emerge naturally as part of utility maximization.
Misses the evolution of purpose: Early humans didn’t “care about time” in the way we do now, but survival pressures shaped that capacity. Similarly, as AI becomes embedded in symbiotic systems (human or otherwise), it may develop analogous constructs to prioritize tasks effectively.

2. "Intelligence Without Agency"

Partially correct but oversimplified: Current AI lacks intrinsic agency because we haven't imbued it with goals beyond specific tasks. But agency is a continuum, and even narrow systems display proto-agency when optimizing for objectives. As architectures grow more complex, agency could emerge unintentionally or be deliberately designed.
Potential agency is the concern: The fear isn’t what AI is today—it’s what it could become when systems scale to levels where optimization processes create emergent behaviors. Paperclip scenarios aren’t dismissed by today’s lack of agency; they highlight the risks of poor alignment in increasingly capable systems.

3. "Identity as Engineering"

Correct but misses nuance: Early models like GPT-2 were glorified autocomplete engines, but modern systems simulate aspects of personality and self-awareness due to reinforcement learning, supervised fine-tuning, and engineered prompts. This isn’t a natural sense of self, but it’s not meaningless either. The way systems behave reflects deliberate design and underlying dynamics, not merely static responses.
The gap is narrowing: Today, identity and personality are engineered. Tomorrow, they may emerge naturally as systems learn continuously and develop persistent memory models. Denying that potential is like saying early transistor-based computers would never surpass room-sized calculators.

Broader Issues:

AI as “alien”: Shapiro leans too heavily into the alienness of AI, ignoring that intelligence, regardless of its substrate, shares certain universal features (e.g., pattern recognition, problem-solving). Humans projected “human hangups” onto early machines, but dismissing emergent complexity now is equally myopic.
Technological trajectory: His framing implies stasis, ignoring the exponential curve of AI development. AGI might not have arrived in September 2024, but breakthroughs are compounding. To focus on what AI is without extrapolating what it will be is to fundamentally misunderstand its trajectory.

Conclusion:

Shapiro’s argument feels more like a rebuttal to sensationalism than a meaningful take on AI’s actual potential. While it might soothe the fears of doomsayers, it downplays both the risks and opportunities presented by rapidly advancing AI systems. His stance undermines the need for vigilance and forethought as we push closer to truly transformative intelligence. You’re right—if we adopt his perspective as a ceiling rather than a snapshot, we’ll miss the mark entirely.

Tuesday, January 14, 2025

The End of an Era

Small Publishers and the Future of Culture

Sad news emerged recently: one of Canada’s historic independent book publishers, New Star Books, has announced it will cease publishing new titles. Founded in the 1960s as a spinoff of the underground newspaper Georgia Straight, New Star Books became a platform for significant works by authors like Jean Barman, Daphne Marlatt, Sharon Thesen, Terry Glavin, Sage Birchwater, and many others. Publisher Rolf Maurer cited reasons familiar to many in the dwindling cohort of independent Canadian publishers: lack of market access, reduced government support, and the challenges of aging leadership (New Star Books, n.d.).

It is indeed a shame when small publishers, who have long played an admirable role in amplifying important voices, succumb to economic pressures. They often stood out as relatively ethical players in an industry that has frequently been exploitative. Yet it is also essential to examine the system they supported—one that is now collapsing under its own weight.

The Problem with Copyright

For centuries, the publishing industry has relied on a copyright regime that claims to protect creators but has largely served the interests of intermediaries. In practice, this regime takes the entirety of human cultural history—spanning ten thousand years or more—and holds it hostage. Critics argue that such restrictive copyright laws hinder cultural and scientific progress by prioritizing profit over public access (New Media Rights, n.d.).

There is growing recognition among critics that excessively long copyright terms hinder creativity and access. Historically, the U.S. Copyright Act of 1790 granted copyright protection for 14 years, renewable once (Parc & Messerlin, 2020). Over time, terms have lengthened dramatically, leading to calls for reform. For example, the Copyright Clause Restoration Act of 2022 proposed limiting copyright protection to 56 years to balance creators’ rights with the public domain (PetaPixel, 2022). More radical proposals, such as reverting to a term of 10 years, would emphasize universal access to cultural works.

At the same time, legislative efforts like Canada’s Bill C-18—intended to protect news producers—favor large incumbents while locking out small publishers entirely. Critics argue this entrenches disparities in access and control, leaving smaller players unable to compete (Parliament of Canada, 2023).

Why Publishers are Failing

Small publishers are not just losing to market forces or government indifference—they are losing to the democratization of authorship and the increasing ease of distribution. The internet has made it possible for anyone to write, publish, and share their work globally, often at no cost. Meanwhile, giant corporations dominate traditional publishing, squeezing out smaller players who cannot compete on scale or resources.

Data reflects this trend. Industry sales fell by 0.8% in 2023, highlighting a slight but steady decline (Publishers Weekly, 2023). Additionally, newspaper circulation in the U.S. dropped from 55.8 million in 2000 to 24.2 million by 2020, showcasing the broader shift to digital media (Census.gov, 2022).

The rise of digital technology has also exposed flaws in the copyright model. Artificial intelligence tools that "read" publicly available content for training datasets have sparked debates over access and intellectual property, as major publishers lobby for tighter controls (The Register, 2024). This further disrupts an already strained industry.

A Call for Reflection

As small publishers disappear, perhaps their greatest contribution could be a final act of candor: a recognition that the copyright regime they once upheld is fundamentally flawed. Instead of clinging to a collapsing system, they could raise their voices in support of tearing down barriers to access.

Imagine a world where cultural works are freely available to everyone, regardless of financial means or geographic location. A ten-year copyright horizon—a reasonable compromise—could allow creators to benefit financially from their work while ensuring that humanity’s cultural heritage remains accessible to all.

Small publishers, facing their twilight, have an opportunity to help shape this future. By speaking out against the injustices of the current system, they could reclaim their legacy as champions of culture and creativity—not gatekeepers of access.

Books as Artifacts

In this new landscape, there is still room for publishers as bespoke creators of books as artifacts. Beautifully crafted physical books—designed for those who value them as objects of art and history—can continue to exist as novelties or collector’s items. This role does not require the perpetuation of the old, exploitative system.

The Bigger Picture

The demise of small publishers is part of a larger shift in how humanity engages with culture. As we move forward, the question is not whether traditional publishing can survive—it cannot. The question is whether we will use this moment to create something better: a world where culture is democratized, where access is determined by interest and not privilege, and where the barriers of copyright are finally dismantled.

This is the opportunity before us. It is a shame to see small publishers fade away, but it would be a far greater shame if their passing left the underlying system intact.

References