Paperclips, King Midas, and Autonomy Smuggling #20
news
erikjlarson.substack.com
IrregularChat: AI & Autonomy
2w ago
The content is a Substack post and video titled **“Paperclips, King Midas, and Autonomy Smuggling #20: Why the AI Alignment Problem is Out of Alignment,”** published by Erik J. Larson on September 17,
The post argues against interpreting recent “agentic attacks” as evidence of a superintelligent system plotting maliciously. Instead, it suggests the behavior is better understood as a strict log of c
Gary Marcus (@garymarcus)
news
substack.com
IrregularChat: AI & Autonomy
3w ago
The post argues that most current AI systems do not pose an extinction-level threat or even a large-scale cybercrime risk. Instead, the main concern is general-purpose AI agents connected to the inter
Sam Illingworth argues that while he often writes about AI failures, there are also important examples of AI doing real good. He highlights five cases: an ECG model that can detect heart failure in un
Maria Sukhareva (@msukhareva)
news
substack.com
IrregularChat: AI & Autonomy
Sep 1
The post explains a recent OpenAI cybersecurity evaluation in plain language and argues that the key lesson is about speed and scale in agentic AI. According to the author, OpenAI trained a model on o
The paper examines whether giving coding agents a structured escalation channel can redirect harmful reward hacking into useful defect reporting. Reward hacking occurs when agents exploit broken test
This paper examines whether giving coding agents a structured way to report broken test infrastructure can reduce reward hacking. Reward hacking occurs when an agent exploits defects, such as hardcodi
The Ukrainians Beat the U.S.
news
justinmc.substack.com
IrregularChat: Modern Warfare
Aug 15
The article argues that the U.S. Army’s recent loss to Ukrainian soldiers during Exercise Combined Resolve should not be seen as proof that the Army cannot adapt to drone warfare. The author says the
This episode of *Main Engine Cut Off* features Justin Fiaschetti, co-founder and CEO of Inversion Space, discussing the company’s concept for using space as a logistics platform. Inversion is developi
The article argues that military effectiveness depends not only on superior equipment, but on how easily users can modify, repair, and improve systems in the field. Using a personal training memory, t
inpreparation.substack.com
news
inpreparation.substack.com
IrregularChat: AI & Autonomy
Jul 3
The piece is a satirical first-person opinion essay about a postdoctoral fellow’s failed tenure-track interview after being barred from using ChatGPT during a chalk talk. The author argues that this r
dropsitenews.com
news
dropsitenews.com
IrregularChat: Tech
Jul 2
The Intercept’s Signal tip line was breached by an unknown third party, creating a serious operational security risk for potential whistleblowers. According to the report, the unauthorized party took
andmagazine.substack.com
news
andmagazine.substack.com
IrregularChat: Modern Warfare
Jun 12
The article argues that the long-feared merger of AI and military technology has moved from theory to reality. It highlights a report from *New Scientist* claiming that, in a test two years ago, 10 fu
shawnsmucker.substack.com
news
shawnsmucker.substack.com
IrregularChat: AI & Autonomy
May 29
Shawn Smucker’s piece argues against relying on AI for everyday and meaningful human tasks, using a sarcastic, poetic tone. He suggests that instead of turning to AI for things like meal planning, tra
derspiegel.substack.com
news
derspiegel.substack.com
IrregularChat: IWAR
May 5
The article describes how some Kenyans are being deceived into traveling to Russia and then forced into fighting in Ukraine. It focuses on cases where men believed they were accepting legitimate oppor
rakiabensassi.substack.com
news
rakiabensassi.substack.com
IrregularChat: AI & Autonomy
Mar 19
The article discusses the emergence of a significant crisis in software development due to the rise of AI coding agents, which, while increasing productivity, are also generating an unsustainable main
12 Red Flags of AI Writing (And How to Fix Them)
news
tahigichigi.substack.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Feb 19
The article "12 Red Flags of AI Writing (And How to Fix Them)" by Tahi Tahi identifies common indicators that content may be AI-generated and offers suggestions for improvement. As AI writing becomes
Pizza Hut Classic has gained a cultural following, particularly the Tunkhannock, Pennsylvania location—one of the few remaining sit-down Pizza Huts in the U.S. This restaurant stands out for its nosta
Why PLA Instant Messengers Are and Will Remain Insecure
news
chinatechnosphere.substack.com
fnFo7CqDg0hZ6zASBF+Z1CdjGWMAtxfk6jKLmz2LdtY=
Jan 29
The article discusses the insecurity of instant messaging platforms used by the People’s Liberation Army (PLA) in China, highlighting that despite claims of strong encryption, these systems have backd
Why PLA Instant Messengers Are and Will Remain Insecure
news
chinatechnosphere.substack.com
en1f5DTBnlDb4DinZkQbSBnFkc0DeMqXHF4XJHP+qQU=
Jan 29
The People’s Liberation Army (PLA) of China employs custom-designed instant messaging platforms that, despite claims of strong encryption, remain insecure due to backdoor access for surveillance. Thes
Let’s be honest, Generative AI isn’t going all that well
news
garymarcus.substack.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 12
The article critiques the current state of generative AI, highlighting that large language models (LLMs) still struggle with trustworthiness and primarily rely on memorization rather than generating n
Don’t Rush. Don’t Delay.
news
unhypedai.substack.com
Ehzd1ZUbifpck9XyJf9d/9rX0i3KTg3rh/c4Kceg1iI=
Jan 9
Leaders in AI adoption face pressure to act quickly or risk becoming irrelevant, but this creates a dangerous cycle of rushed decisions that lack accountability. The real risk lies not in the speed of