[ExI] Bernie Sanders and his stupidity preservation bill

[BillK]

On Fri, 4 Sept 2026 at 10:37, John Clark via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> *Bernie Sanders has proposed a bill that would ban superintelligence
> research, give a "corporate death penalty" to any company that violates the
> ban, and put in prison for 20 years any individual who engages in
> superintelligence research.  Chinese leaders must be smiling. *
> <snip>
> _______________________________________________
>


I asked Qwen AI to review two articles about this proposal:
Joe Rogero (from MIRI-(If Anyone Builds it, Everyone dies) and Gary Marcus.
MIRI want a pause to AI development, whiie Gary worries about China
continuing development.
I thought Qwen did a good review of the problem.
BillK

Question:
Can you review and discuss these two articles? -
https://aistop.watch/p/shutting-it-down?open=false#%C2%A7sanders-and-casar-propose-superintelligence-ban
and https://garymarcus.substack.com/p/the-new-sanders-casar-ban-artificial


Qwen3.8-Max Thinking -

Both articles, published on September 3, 2026, address a major breaking
event in AI policy: the introduction of the *Ban Artificial
Superintelligence Act* by Senator Bernie Sanders and Representative Greg
Casar. The proposed legislation seeks to outlaw the development and
deployment of AI systems that match or exceed human cognitive performance,
halt "advanced AI development" until federal standards are met, establish a
cabinet-level AI regulatory agency, and push for a global treaty to enforce
these rules.
While both authors recognize the severe dangers posed by advanced AI and
applaud the government's urgency, they reach dramatically different
conclusions about whether the legislation is a necessary lifeline for
humanity or a naive policy misstep.
Here is a review and discussion of both perspectives.
1. AI StopWatch: "Shutting it down" (by Joe Rogero)
*Stance:* Enthusiastically supportive. Joe Rogero views the Sanders-Casar
bill as a "major policy breakthrough" and a "massive breakthrough in sane
AI policy." He writes that the news moved him almost to tears, framing the
legislation as a vital step toward ensuring human survival.
*Key Arguments:*

   - *The Precautionary Principle:* Rogero argues that it is common sense
   not to build something smarter than ourselves that we cannot control. He
   strongly supports the hard ban on superintelligence and AI systems capable
   of planning the "disempowerment of humanity."
   - *Unilateral Leadership:* While acknowledging that a unilateral U.S.
   halt isn't a permanent fix without international cooperation, he views a
   bold U.S. stand as the best way to force countries like China to treat the
   issue with the gravity it deserves, akin to nuclear non-proliferation
   treaties.
   - *Accepting Bluntness:* He admits the definition of "advanced AI
   development" might be overly broad and the international policy is
   currently thin, but he believes that erring heavily on the side of caution
   is the only rational choice when no serious regulations currently exist.

*(Note: The broader issue of AI StopWatch also covers the bipartisan "Stop
Rogue AI Act," which mandates technical security standards for AI agents, a
survey showing that 111 national security experts are deeply concerned
about AI extinction risks, and NYC's one-year ban on classroom AI).*
------------------------------
2. Gary Marcus: "The new Sanders-Casar Ban Artificial Superintelligence Act
– and why I oppose it"
*Stance:* Pragmatically opposed to the bill's specifics, while supportive
of its underlying regulatory goals. Cognitive psychologist and AI critic
Gary Marcus agrees with the need for a temporary pause and a federal AI
agency (a stance he has held since 2023). However, he strongly opposes the
bill as written, arguing it goes "too far."
*Key Arguments:*

   - *Permanence and Geopolitics:* Marcus argues that a permanent,
   unilateral ban on superhuman AI is too broad. He warns it is a "guarantee
   of leaving the US behind," creating a scenario where authoritarian regimes
   will simply ignore the ban and gain a massive strategic advantage.
   - *Benchmarking Naivety:* He notes that the bill relies on complex and
   often flawed benchmarks to define "superintelligence," making a permanent
   legal ban difficult to enforce accurately.
   - *Sci-Fi vs. Reality:* Aligning with virologist Dr. Angela Rasmussen,
   Marcus criticizes the bill for focusing entirely on hypothetical,
   existential "sci-fi" risks (like AI disempowering humanity) while ignoring
   the immediate, real-world harms AI is causing right now (such as cyber
   vulnerabilities, disinformation, and algorithmic bias).
   - *A Conditional Alternative:* Instead of a permanent ban, Marcus
   proposes borrowing language from physicist Anthony Aguirre: a conditional
   halt. Development of highly general and autonomous AI would only be
   permitted if companies can demonstrate with "high assurance" to an
   independent regulatory authority that sufficient alignment and safety
   controls exist. This shifts the burden of proof onto the tech companies.

------------------------------
Discussion & Synthesis
These two articles perfectly encapsulate the current ideological divide in
AI safety and policy discourse. They share several foundational beliefs but
diverge sharply on execution.
*The Common Ground*
Both authors agree on three vital points:

   1. *The Era of Laissez-Faire is Over:* Both agree that private tech
   companies can no longer be trusted to self-regulate frontier AI models.
   2. *The Need for a Federal Agency:* Both strongly support the creation
   of a dedicated, cabinet-level federal body to oversee AI, enforce
   standards, and monitor the lifecycle of frontier models.
   3. *The Danger of Proliferation:* Both are deeply concerned by recent
   events (such as the Hugging Face attack mentioned in both articles) and the
   rapid release of open-weights models with dangerous cyber capabilities.

*The Geopolitical Prisoner's Dilemma*
The most fascinating divergence is how the two handle the "China factor."

   - *AI StopWatch* hopes for a "moral contagion" effect—that a brave,
   unilateral pause by the United States will shame or diplomatically force
   rival nations into a global non-proliferation treaty.
   - *Marcus* views this through the lens of realpolitik. He sees a
   unilateral ban as unilateral disarmament. If the U.S. makes advanced AI
   research a criminal offense punishable by 20 years in prison, top talent
   will either flee to jurisdictions without such bans or work in secret,
   handing a massive geopolitical advantage to adversarial nations.

*The Burden of Proof: Hard Bans vs. High Bars*
Marcus’s proposed alternative (borrowed from Aguirre) offers a highly
pragmatic middle ground that the Sanders-Casar bill lacks. Instead of
Congress trying to permanently outlaw a moving target (superintelligence),
Marcus suggests establishing a regulatory "high bar." If a company wants to
build a system smarter than humans, *they* must prove to a federal agency
that it is safe and aligned. If they cannot prove it, development halts.
This incentivizes safety research and leaves the door open for beneficial
innovation, whereas AI StopWatch embraces a permanent, blunt instrument to
guarantee survival.
*X-Risk vs. Immediate Harms*
Finally, the articles highlight a philosophical debate about what we are
actually trying to regulate. The Sanders-Casar bill—and AI StopWatch’s
support of it—is rooted in *Existential Risk (X-Risk)*. The primary fear is
a rogue, superintelligent agent destroying human civilization. Marcus,
however, represents the *AI Ethics/Immediate Harm* camp. He argues that by
drafting sweeping legislation around hypothetical "Terminator" scenarios,
lawmakers are failing to address the very real algorithmic damages
happening today.
Conclusion
The Sanders-Casar "Ban Artificial Superintelligence Act" clearly represents
a watershed moment in how Washington views AI. AI StopWatch is right to
celebrate the fact that the existential threat of AI is finally being
treated with the legislative gravity it warrants. However, Gary Marcus’s
critique highlights the practical flaws of reactionary legislation. Moving
forward, the most effective policy will likely need to marry the aggressive
urgency of the Sanders-Casar bill with the conditional, burden-shifting
regulatory frameworks that Marcus advocates.
--------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20260904/e3efe310/attachment-0001.htm>