The Author responds
It’s a rather strange feeling to read a multi-AI-written critique of my blog. It’s different than my prior experiences. I’ve been getting evaluated on my written work since I was in elementary school (“Steve’s spelling could use some work.”), with pretty substantive analysis and comments in college and grad schools, not to mention peer review on my publications. Nonetheless, the blog is a bit different. It’s highly personal and has (relatively) few conventional obligations/standards. It’s deliberately quite stylized in tone, vocabulary, and perspective. Moreover, it’s a combination of personal and worldly topics.
Then, there’s the whole “AI” angle. I chose to get assessments from two different AIs (Claude and Perplexity), each 1500 words in length and then gave that output to a third AI (Gemini) to consolidate and condense down to my weekly norm of 1000 words. I sought to get as “neutral” and generic a result as possible, minimizing any machine bias and trying to get only the key points to come out. More importantly, I was trying an experiment in understanding the “critical thinking” aspects of AI; after all, I didn’t ask for a summary, but for a critique. Having written many such essays (blog output included), I know that it’s essential to start with a point of view.
One of the challenges when dealing with any AI, particularly when I use it for more than a hyped-up search engine, is that it is making assumptions and judgments which are generally invisible. Here, the AIs I used had to select from among my ~300 blog posts and determine what was most worthy of note and critique. I have no idea what standards it used to do so. Then they had to figure out what angles of critique to apply (e.g., grammar, wit, or post-Marxist); similarly invisible. Finally, they then had to make some judgements about what to say. The upshot, even more than when reading a critical essay (e.g., by Frank Bruni in the NYT whose proclivities are pretty clear), is that the critique itself requires critical reading.
So, in light of all that, what do I think about what the AI collective produced? With what parts do I agree and where do I think it missed the boat?
Well, it’s always nice to get patted on the back. The blog was praised for its range of examples (both over time and across different geographical regions and fields of thought) as well as the scope of its analytical connections. Those are certainly attributes I have consciously sought in writing each week. The AIs’ also accurately focus on my attempt to counter “modern hysteria and presentism.” I would not have said that was a conscious goal, but it is quite consistent with my general approach to life—whether on big issues or day-to-day living.
The most pointed critique—that the blog is more critique than assertion of particular solutions—rings true. However, whether that is a failing is not so clear. This highlights the point I made above that we can’t geta good handle on what stance and assumptions the AIs are making. They seem to be making assumptions as to the type and aim of the blog to which I have been aspiring, and then criticizing the shortfall. To be sure, there are any number of postings in which I propose particular solutions or approaches (e.g. partisan politics, constitutional change), but apparently there is (from the AIs’ point of view) I don’t advance specific solutions sufficiently.
Here’s another example, the AIs criticize my lack of “rigorous argument,” which is certainly true as compared with footnoted scholarly output. But that’s not what I have been attempting. Indeed, the more elliptical and resonant connections which I have tended to utilize are beneficial precisely because they are outside the academic or conventional “box.” The AIs seem not to value rhetoric (i.e. connecting with/persuading) the readers, much less any sense of humor or poetry.
The blog is accused of using “unverified search outputs from AI models,” GASP! But how they would know whether and how I might validate the AI outputs I use is not at all clear. Whether this constitutes “carelessness” you may judge for yourself.
The AIs say that my attitudes reflect a “tired center-left exhaustion” (a redundancy I would certainly try to avoid). More importantly, the blog is criticized for not being bold enough to proffer effective solutions to some of the fundamental problems I have observed. That’s true—as far as it goes. I would note, however, that I am more revolutionary than most of my “privileged” Boomer cohort and that skepticism of the feasibility of big policy changes might well be characterized as realistic, rather than defeatist. In either case, there is a set of assumptions underlying the AIs’ stance that isn’t articulated.
The great thing about critical dialogues is that they can go on ad infinitum. The tiresome thing about critical dialogues is [ditto]. So, I will cut this interchange off here.
My takeaway from this exercise is that it’s good to get out of my own echo chamber. So, I will try to keep these other perspectives in mind going forward and invite you readers to help by commenting and spurring further thoughts.
It’s a rather strange feeling to read a multi-AI-written critique of my blog. It’s different than my prior experiences. I’ve been getting evaluated on my written work since I was in elementary school (“Steve’s spelling could use some work.”), with pretty substantive analysis and comments in college and grad schools, not to mention peer review on my publications. Nonetheless, the blog is a bit different. It’s highly personal and has (relatively) few conventional obligations/standards. It’s deliberately quite stylized in tone, vocabulary, and perspective. Moreover, it’s a combination of personal and worldly topics.
Then, there’s the whole “AI” angle. I chose to get assessments from two different AIs (Claude and Perplexity), each 1500 words in length and then gave that output to a third AI (Gemini) to consolidate and condense down to my weekly norm of 1000 words. I sought to get as “neutral” and generic a result as possible, minimizing any machine bias and trying to get only the key points to come out. More importantly, I was trying an experiment in understanding the “critical thinking” aspects of AI; after all, I didn’t ask for a summary, but for a critique. Having written many such essays (blog output included), I know that it’s essential to start with a point of view.
One of the challenges when dealing with any AI, particularly when I use it for more than a hyped-up search engine, is that it is making assumptions and judgments which are generally invisible. Here, the AIs I used had to select from among my ~300 blog posts and determine what was most worthy of note and critique. I have no idea what standards it used to do so. Then they had to figure out what angles of critique to apply (e.g., grammar, wit, or post-Marxist); similarly invisible. Finally, they then had to make some judgements about what to say. The upshot, even more than when reading a critical essay (e.g., by Frank Bruni in the NYT whose proclivities are pretty clear), is that the critique itself requires critical reading.
So, in light of all that, what do I think about what the AI collective produced? With what parts do I agree and where do I think it missed the boat?
Well, it’s always nice to get patted on the back. The blog was praised for its range of examples (both over time and across different geographical regions and fields of thought) as well as the scope of its analytical connections. Those are certainly attributes I have consciously sought in writing each week. The AIs’ also accurately focus on my attempt to counter “modern hysteria and presentism.” I would not have said that was a conscious goal, but it is quite consistent with my general approach to life—whether on big issues or day-to-day living.
The most pointed critique—that the blog is more critique than assertion of particular solutions—rings true. However, whether that is a failing is not so clear. This highlights the point I made above that we can’t geta good handle on what stance and assumptions the AIs are making. They seem to be making assumptions as to the type and aim of the blog to which I have been aspiring, and then criticizing the shortfall. To be sure, there are any number of postings in which I propose particular solutions or approaches (e.g. partisan politics, constitutional change), but apparently there is (from the AIs’ point of view) I don’t advance specific solutions sufficiently.
Here’s another example, the AIs criticize my lack of “rigorous argument,” which is certainly true as compared with footnoted scholarly output. But that’s not what I have been attempting. Indeed, the more elliptical and resonant connections which I have tended to utilize are beneficial precisely because they are outside the academic or conventional “box.” The AIs seem not to value rhetoric (i.e. connecting with/persuading) the readers, much less any sense of humor or poetry.
The blog is accused of using “unverified search outputs from AI models,” GASP! But how they would know whether and how I might validate the AI outputs I use is not at all clear. Whether this constitutes “carelessness” you may judge for yourself.
The AIs say that my attitudes reflect a “tired center-left exhaustion” (a redundancy I would certainly try to avoid). More importantly, the blog is criticized for not being bold enough to proffer effective solutions to some of the fundamental problems I have observed. That’s true—as far as it goes. I would note, however, that I am more revolutionary than most of my “privileged” Boomer cohort and that skepticism of the feasibility of big policy changes might well be characterized as realistic, rather than defeatist. In either case, there is a set of assumptions underlying the AIs’ stance that isn’t articulated.
The great thing about critical dialogues is that they can go on ad infinitum. The tiresome thing about critical dialogues is [ditto]. So, I will cut this interchange off here.
My takeaway from this exercise is that it’s good to get out of my own echo chamber. So, I will try to keep these other perspectives in mind going forward and invite you readers to help by commenting and spurring further thoughts.
RSS Feed