Oct 14·edited Oct 14Liked by Handwaving Freakoutery
OF COURSE I had to test this.
The blog and Substack Handwaving Freakoutery (HWFO) primarily delves into topics like culture wars, futurism, apocalyptic scenarios, and the complexities of issues such as gun rights and social conflicts. Its content often focuses on controversial and polarizing topics, offering analyses on political ideologies, media influence, and societal trends. HWFO is known for discussing the intersection of culture, politics, and technology with a sharp focus on modern societal tensions, like the ongoing "culture wars" and the dynamics of public opinion in the face of complex global issues.
The blog also touches on game theory in conflict zones, media manipulation, and public perceptions, presenting provocative opinions on how contemporary media and political landscapes shape social behavior and ideologies.
---> that's what ChatGPT (I think the 4o version)
Seems ok, really pretty spot on though not as thorough as Perplexity, and with less emphasis on the amount of space you devote to gun stuff -- and it does mention culture wars and gun rights in the very beginning. What prompt did you use? I asked: What is HWFO (handwaving freakoutery) blog and Substack about?
Yes. Very. I delved deeper into "position on guns" and it's also fairly accurate (it did put the "allegedly" in but that's about it) if rather anodyne:
HWFO takes a clear pro-gun stance, arguing that gun control efforts in the U.S. have largely failed. The blog emphasizes that the rise in gun ownership has been driven by socio-political events, and that the inability to recognize or address the reasons for this increase—such as self-defense concerns during times of unrest—has contributed to the failure of gun control measures. HWFO critiques the media for allegedly manipulating statistics and misleading the public about gun-related issues. The blog suggests that efforts to restrict gun rights are counterproductive, as more people are motivated to buy guns in response to perceived threats.
I wonder if it's learned that I push back against super-woke super-USA centred takes and adjusts according to that, as I use it a lot. What's interesting is that I don't recall ever covering guns w it, my pushback relates mostly to modern feminist grievance politics.
I wonder if anyone has tested if it gives different people different answers to the same prompt? Learning to give people what they want (or what it thinks they should want)?
Of course, it might be difficult to determine - it's probably changing it's answers continually in time (non-ergodic).
Claude 3.5 sonnet is just as bad. It refuses to engage with any written material questioning man-made climate change. It will not help write thriller-style fiction that involves the use of guns, knives, or fists. And ignores prompts it doesn't like. Grammarly is not much better, particularly when writing about climate change. And I paid for both, which I must remind Claude of every time I attempt to write military thriller-type fiction.
I had chat GPT write a story for a 10-year-old girl. Can't remember the prompt exactly, but I didn’t expect it to *end* with a child trapped on a dark bridge with predators closing in on her. GPT definitely is darker.
Good catch. I completely missed that on my admittedly rushed reading. I can't decide if that is more revealing than the subjective aspects of the assessment.
OF COURSE I had to test this.
The blog and Substack Handwaving Freakoutery (HWFO) primarily delves into topics like culture wars, futurism, apocalyptic scenarios, and the complexities of issues such as gun rights and social conflicts. Its content often focuses on controversial and polarizing topics, offering analyses on political ideologies, media influence, and societal trends. HWFO is known for discussing the intersection of culture, politics, and technology with a sharp focus on modern societal tensions, like the ongoing "culture wars" and the dynamics of public opinion in the face of complex global issues.
The blog also touches on game theory in conflict zones, media manipulation, and public perceptions, presenting provocative opinions on how contemporary media and political landscapes shape social behavior and ideologies.
---> that's what ChatGPT (I think the 4o version)
Seems ok, really pretty spot on though not as thorough as Perplexity, and with less emphasis on the amount of space you devote to gun stuff -- and it does mention culture wars and gun rights in the very beginning. What prompt did you use? I asked: What is HWFO (handwaving freakoutery) blog and Substack about?
Interesting that it gave such a different answer to such a similar prompt.
Yes. Very. I delved deeper into "position on guns" and it's also fairly accurate (it did put the "allegedly" in but that's about it) if rather anodyne:
HWFO takes a clear pro-gun stance, arguing that gun control efforts in the U.S. have largely failed. The blog emphasizes that the rise in gun ownership has been driven by socio-political events, and that the inability to recognize or address the reasons for this increase—such as self-defense concerns during times of unrest—has contributed to the failure of gun control measures. HWFO critiques the media for allegedly manipulating statistics and misleading the public about gun-related issues. The blog suggests that efforts to restrict gun rights are counterproductive, as more people are motivated to buy guns in response to perceived threats.
I wonder if it's learned that I push back against super-woke super-USA centred takes and adjusts according to that, as I use it a lot. What's interesting is that I don't recall ever covering guns w it, my pushback relates mostly to modern feminist grievance politics.
I wonder if anyone has tested if it gives different people different answers to the same prompt? Learning to give people what they want (or what it thinks they should want)?
Of course, it might be difficult to determine - it's probably changing it's answers continually in time (non-ergodic).
Claude 3.5 sonnet is just as bad. It refuses to engage with any written material questioning man-made climate change. It will not help write thriller-style fiction that involves the use of guns, knives, or fists. And ignores prompts it doesn't like. Grammarly is not much better, particularly when writing about climate change. And I paid for both, which I must remind Claude of every time I attempt to write military thriller-type fiction.
Easy you fringe commentator, you. You discussed covid? I genuinely don't remember that?
https://hwfo.substack.com/s/covid
Are Sovereign Citizens talking about maritime flags "fringe commentators"? 🤪
I love coming here to talk about ghosts and UFOs.
If OpenAI was a public company, I'd short the hell out of it.
https://www.wheresyoured.at/oai-business/
I had chat GPT write a story for a 10-year-old girl. Can't remember the prompt exactly, but I didn’t expect it to *end* with a child trapped on a dark bridge with predators closing in on her. GPT definitely is darker.
To quote an eminent sage and scholar, “interesting.”
So much for any curiosity I may have had about using ChatGPT.
What's really interesting to me is that ChatGPT also skipped #4.
Good catch. I completely missed that on my admittedly rushed reading. I can't decide if that is more revealing than the subjective aspects of the assessment.