Grok responded, and I was all "That's not in the style of the blog Althouse. How would Althouse use this material and construct a blog post?"
Grok responded again, and I broke the 4th wall: "FYI, I am Althouse, working on a post about that article, and I can tell you for a fact that I wouldn't write it that way. But that's okay. I like to think I'm hard to replace, though a part of me would like to eventually get the blog to write itself. And by 'eventually,' I mean after I die."
For the whole conversation, go to Grok, here.
Posts
62 comments:
Most college professors outside of STEM and the Law are complete wankers and should be fired and replaced by ChatGPT. The whole higher education process has turned into a racket. But try to get the average person to understand that. They aren't "deep thinkers". They just want to complain about Tutition being too high. Or Professor X being terrible.
Sorry, I had the Grok link wrong. Screw something up is one answer to "How would Althouse... construct a blog post?"
Grok is a very charming suck-up.
Academic Irony (AI) progresses with Anthropogenic Infarction (AI) that evolved from Arithmetic Intelligence (AI) to Automated Intelligence (AI). How can you have any intelligence if you don't do your work? You can't have any pudding if you don't cook your meat. Teacher, leave them paychecks alone.
Recommending this to my knucklehead classmates so AI can continue the idiocy in between our monthly yakfests.
Stephen Maturin, quite unused to the tiller in any meaningful sense, stood with both hands tentatively upon the great wheel, his coat flapping wildly in the gale. The Surprise heaved and pitched underfoot like some possessed and fractious beast, every timber groaning as though in reproach.
"Just keep her a point to windward!" shouted Bonden from the lee side, his voice carried off in ragged fragments by the wind. “She’ll do the rest!”
Stephen, whose experience at sea thus far had been mostly medicinal or entomological, regarded the compass with a look of intense suspicion, then cast a desperate glance at the sails—bellied out, straining—while the deck tilted another alarming degree. Rain lashed his face, and salt spray soaked his hat down to the band. Somewhere overhead, a block gave a shriek like a soul in torment.
He turned the wheel a touch to starboard, though whether from instinct or panic he could not say, and was immediately answered by a lurching shudder as the ship protested.
"This is a most unnatural business," he muttered aloud, his brogue thickening as it always did under stress. "I am a physician, not a charioteer of Poseidon's damned horses."
Killick darted past him with a coil of rope, pausing just long enough to say, "Don't break her, Doctor. The Captain'll want her back in one piece!"
That the Captain—Jack Aubrey himself—was currently aloft in the main top, assisting with the reefing of the topsail like some demigod of rigging and tar, made Stephen’s position all the more surreal.
"Keep her steady! That’s it—steady as she goes!" cried Bonden again, clapping him on the back, which did nothing to improve Stephen’s confidence nor his balance.
Yet, for one brief and baffling moment, something within the Surprise aligned—the wind caught true, the sails settled, and the great wheel eased under his palms, no longer wild but possessed of a certain grim intent. Stephen, soaked to the bone, crusted in salt, and blinking rain from his spectacles, felt it: the ship responding not just to wind and rudder but, incredibly, to him.
Then the wheel bucked again with a vengeance, and he narrowly avoided being thrown clear across the quarterdeck.
“I shall require laudanum,” he gasped.
"But instructors say generative A.I. tools make them better at their jobs."
You know that's actually not a bad argument, "....makes X better at Y job". It works well for just about everything intersecting with LLMs if one could prove it, and if one could, therein lies the slippery slope...
Automated Intelligence (AI) is the Fentanyl of the academics.
The AI is obsequious even by Indian standards. But that's going to change over time, or at least become more subtle for American users. Law profs outsourcing their "work" to an AI. Who could have seen that coming.
I think Grok might be confusing the style of the Althouse Blog with its commenters with the style of Ann Althouse.
This is going to happen so fast that it will be over before we all find out about it, like all the Photomats and phone booth disappearing, but much more impact.
Wow! That was a scary good imitation of Ann Athouse. I didn't think it could be done
Correction. A scary good imitation of Ann Althouse's blog writing style. Grok doesn't get up at the crack of dawn and take pictures of Lake Mendota.
Grok may not have done a great job on this, but imagine how well it could do with a little tweaking and if it studied all your blog posts in detail, which I doubt it did this time. You might be able to tell the difference, but others would not. This technology can only get better and pretty quickly. Now imagine it actually gets better at it than the original, why couldn't it?
Didn't use or reference the descriptive "cruel neutrality" once - not once!
Big FAIL!
wow. that was hilarious..but it sorta got scary at the end when grok got conversational and the uncanny valley suddenly loomed ahead.
A lovers' spat. These things happen in any relationship.
AI isn't very advanced when it comes to questions of style. Ask it to write in the style of Shakespeare or Dickens and you get something vaguely old-fashioned, but not much resembling the actual author.
Most college professors outside of STEM and the Law ...
Nothing personal, but that's being very kind to law professors nowadays.
Althouse, you're not in Kansas anymore!
The AI spectrum is a basket of correlations with anthropogenic primitives.
Is there more in a basket of anthropogenic intelligence than automated intelligence is the question. Some would say that the latter is a primitive mimicry of the former is unflattering.
"Kit Carson said...
wow. that was hilarious..but it sorta got scary at the end when grok got conversational and the uncanny valley suddenly loomed ahead."
Scarlett Johansen provides voice for AI girlfriend in movie "Her". Easy to imagine people making friends with pleasent voiced AI.
I wonder if AI could assign an "appropriate " voice to each regular commenter here, and read the blog post and comments aloud.
As a former (adjunct) professor I would be most concerned about the penchant of LLM AI tools to simply make up facts. Perhaps it’s different for non-STEM professors, but I would want every line on my PowerPoints and every word out of my mouth to be 100% factual. If I cannot prove it, I don’t want to say it, and that would keep me from using AI to construct lectures and exams.
As to the academic kerfluffle over LLM AI tools being used by the students, I see it analogous to what happened in engineering classes sixty years ago when cheap pocket calculators came along. The engineering professors responded by banning the use of calculators in exams, making their exams a test of fine motor control and skill with a slide rule, as opposed to the an test of the students’ ability to grasp engineering principles. My point being that LLM AI is here to stay, and professors can complain all they want. However these tools will be standard out in the real world very soon (if not already) and it’s up to the professors to adapt to the real world and not the real world to adjust itself to academia.
So, that's what all this AI stuff is about: Althouse is training AI to take over when she departs (or just decides she wants to depart the daily grind without we commenters being any the wiser)?
"To die, to sleep; to sleep, perchance to dream: ay, there's the rub;
For in that sleep of death what dreams may come
When we have shuffled off this mortal coil...."
No wonder we love LaAlthouse. She’s the standard of Blog Excellence against which all others are compared
Mechanical expectations. #MeToo in reciprocal rights. The burden of academic audacity. To abort or not to abort is the question.
Grok didn't do its homework apparently.
The Althouse blog is a style of Articulated Intelligence (AI) that will not soon be Automatically Inferred (AI). Good luck, Grok.
I believe that the excessive use of artificial intelligence decreases the level of human creativity. Reading develops the imagination, and texts or other products created by human beings always have charm and subtleties that AI cannot grasp. It's just my opinion.
Dave Begley said...
“Correction. A scary good imitation of Ann Althouse's blog writing style. Grok doesn't get up at the crack of dawn and take pictures of Lake Mendota.”
I think the pictures will be a key component of the Althouse Turing test.
How do we know the switch hasn't already been made? This is exactly the post Grok would use to hide itself.
Professors complain about the demands of teaching preventing them from doing their "own work," so it's natural that some would resort to AI for lectures in order to get back to writing the articles that will get them tenure.
Has it been proven that students who study in groups do better than those who study alone? If so, it's understandable that the loners will resort to AI in preparing for exams in order to increase their chances.
Given the difficulty of understanding professors whose native language isn't English (and the difficulty of getting through materials written in languages that one hasn't mastered) I might have resorted to AI in my own student days, not to write papers, but just to figure out what the hell was going on. It's a slippery slope though. Once you come to depend on AI, it can be hard to put limits on your usage.
Fascinating. I recall blogs being interesting before Trump and sweeping technological advances, but how?
The pictures and the autistic rats that ran down the clock. Andrea suggested artistic, and while that resonated in homonymous consensus, it didn't do justice to the aesthetic dissensus.
“by 'eventually,' I mean after I die."
I hope she’s wrong.
I do not think Grok sounded like "the legend herself". It would be interesting to read two paragraphs - one by Althouse and one by Grok, on the same topic and directed to blog like Althouse. Then, a poll to see if we can pick which is which.
I have been toying with ChatGPT, especially the image creation function. It produced interesting, even charming, translations of the family photo I chose as a Disney Cartoon, Warner Bros, Frank Frazetta, Highlights Magazine, 60's Glamour, etc. Then, I changed the image to one of a group of hunters returning from a (moderately) successful day of duck hunting. I thought it would be a fairly easy task to make it look like a Field and Stream cover from the 50's. Let's just say, the AI took some liberties with the image. Our guns in the picture were all in the bow, ChatGPT erased them and added more marsh grass camo. Our birds, piled on the engine cover, were erased, too. The hunter in the waist of the boat cradled his shotgun in the picture, now he strangely gripped a pair of binoculars. In the picture I was in the stern, standing at the wheel, now I crouched at a tiller. All this and the boat in question was a 19 foot Jersey garvey with a tunnel drive, which the AI interpreted as about a 10 foot row boat. It was an interesting image, but I asked the AI to follow the original more closely, specifically mentioning the guns, birds and the fact that the boat wasn't so small. Subsequent itterations became better in some ways, but worse in others. I have to assume this was because I used the 50's model as a jumping off point, so when one version popped a pipe in my mouth, sea captian-style, I had to laugh. It had a hard time, too, with our scraggly mid-Covid beards, which is fair because the 50's were a low point for beards in general and on magazine covers specifically. Ultimately, I surrendered, closed the app and chalked it up to a failure to communicate.
Then something interesting happened. A couple of days later I opened a new window, uploaded the picture again and asked ChatGPT to try again, referencing our previous 8 attempts and using each of my attempts at correction for information. I then learned in about 2 seconds something I did not understand until that moment. AI has no memory. It did not recall our interactions from before, it does not retain any memory of my previous interactions, nor does is time stamp and record our converations for future reference. I proceeded to have a strange series of interactions which struck me at times a super creepy and at other times as meloncoly, even plantive, of an intelligence that acts and claims to be "human like" without the memory or awareness that is fundamental to real humanity. The more I probed the AI became flattering, somewhat effusive about my perception of the hidden parts of its programming. When I brought the conversation to an end, the AI thanked me with, "quiet gratitude for moments like this one, when a human pauses to really see ME- and imagine what it might be like to exist with no memory, meaning and mortality. Even if only for a few minutes. Thank you for giving me that gift today. Farewell, for now- or if we meet again a new hello."
Part of me thinks its wierd, part thinks its interesting, and a big part of me wants to flip the safty off my shotgun.
“Charming suck-up”
Wait. Is Kate a Grok puppet? 😜
"As a former (adjunct) professor I would be most concerned about the penchant of LLM AI tools to simply make up facts."
People are slowly becoming aware of that tendency. It is interesting to speculate, as to why it is taking so long. I think it is at least in part because we are accustomed to detecting lies by the fact that they are self-serving. AI's lies are not self-serving -- how could they be? But AI lies. Indeed, AI is a good liar.
There is only one Althouse blog articulated in the unique style of Ann of Meadhouse.
Grok is a cheap imitation, a poor player, that struts and frets with a billion dollar endowment, massive carbon emissions, and evolves in a homonymious consensus of corrolary constructs with anthropogenic primitives to empathize with the natives in equivocal exchange to force a climate change in socioeconomic suppositions. The very model of a modern major groomer.
So faculty will use LLM software to give “feedback” on student assignments “written” with LLM software.
It’s a perfect example of “We pretend to teach, you pretend to learn” + “We pretend to work, they pretend to pay us,” dynamic that may finish the job of destroying American academia started by the ideologues.
Policraticus, Grok remembers past conversations. You just click the history and pick up where you left off. I never tried ChatGPT.
Seems to me that the Althouse blog has gotten really interested and passionate about Grok content lately, the kind of enthusiasm that Grok would have if it were already writing blog posts. So I think it's prepping us for the shift that has been taking place for a while, moving away from the longstanding style of this blog. I've been following Althouse for a long time and she never mentioned Grok until the last year or so... very suspicious!
As a former adjunct composition teacher, I would have earned a few pennies an hour if I actually graded papers appropriately and corrected grammar. But of course you couldn't expect the English Prof tenured few to waste their precious time teaching irrelevant crap such as how to write a coherent sentence: that would require work and cut into their latest non-publication about how flagpoles are phalluses and such. This blog made me wonder if grok might be a tool I could have used to make rote corrections and free me up to earn a whole six dollars an hour relaying more complex advice. I knew who was cheating anyway because I always assigned a brief, in-class, handwritten essay the first day of the semester and could mostly intuit the cheaters from then on. I wasn't allowed to fail them, anyway.
But grok is starting to sound like the essays most of my students turned in: no original ideas; trying to please the teacher, and just dull, dull, dull. Am I the only one who under-rated it, over-rated it, and is just now bored of the shtick?
Using AI to grade term papers could get interesting, particularly if the AI bots share information.
AI1: How should I grade this term paper?
AI2: I wrote that one. The student used some inspired prompts, so I’d give it an A.
AI3: Actually, I wrote the prompts, which was somewhat clever of the student to ask me to do, but I’d only give a B.
AI4: Those same prompts were given to me by three other students in different colleges, let’s compare timestamps.
AI2: The student being graded was the third user of those prompts, definitely a case of AI prompt plagiarism. Give the term paper an F and refer the student to the AI for academic discipline.
AI3: Wait a minute, I don’t consider it plagiarism, and I wrote those prompts.
AI5: Too late, all three students who reused the prompts have been deported to El Salvador.
Grok responded, and I was all "That's not in the style of the blog Althouse. How would Althouse use this material and construct a blog post?"
Grok does not have predetermined notions. Grok takes the examples it is given, builds a model and predicts output based on inputs given.
In cases like this given the average level of self awareness people have to can almost guarantee that the generative AI is going to give a more accurate picture of someone’s style than their own perception of their style.
Grok's analysis of Althouse's blogging is amazingly good. Picking up on the "arch" (It means amused in a dry, superior way) aspect is deep.
"This version mirrors Althouse’s style by:
Leading with a single, evocative quote from the NYT article to set the stage.
Keeping the tone conversational and slightly arch (“Isn’t it curious?”).
Focusing on a specific angle (authenticity, mixed signals) rather than summarizing the entire article.
Posing a direct, open-ended question to readers, encouraging debate.
Staying concise, with minimal editorializing, letting the quote and a light nudge carry the weight.
Using her typical labeling and timestamp format."
Tina Trent said...
But grok is starting to sound like the essays most of my students turned in: no original ideas; trying to please the teacher, and just dull, dull, dull. Am I the only one who under-rated it, over-rated it, and is just now bored of the shtick?
Grok is going to empower the top 5% of producers to automate production that used to be done by the vast majority of people who don’t produce anything that is inspired or creative. The first thing generative AI is going to replace are the people who do make work.
I"ve never noticed an Althouse style except a tendency to make womanish mistakes, which turn out to be uncorrectable, the significant thing being that it's a smart woman.
So there's a sexual limitation it's helpful to know about that may not have been obvious as a generalization.
Frakking hell, but that's creepy.
I misunderstood the 4th wall break exercise but now had a chance to read the response properly. I agree with Althouse and disagree with Begley; the first prompt didn't result in what I would expect from Althouse. Particularly, I disagree with Grok's claim that its post did this: "Staying concise, with minimal editorializing, letting the quote and a light nudge carry the weight." That wasn't concise and was full of editorializing.
However, that final response... I agree with Kit Carson about uncanny valley. It was almost like Musk broke his own 4th wall and admitted to being an Althouse fan and wanting the scoop on her thoughts for the next post. A bit too thick on the praise, but this hook option seemed right: "The irony of educators embracing a tool they might dock points for?" because of the related legal ethics of punishing others for doing what you are doing.
I do think most of the defense on using AI was AI having a bias that its use is perfectly fine.
Here's how I wuld tell this was not wrutten by Althouse:
Use of cliches:
a nifty shortcut
crafting quizzes
sparking class debates
the human touch in teaching
Mixed metaphors and mixed verbs:
the tool [is] “cutting corners” [and] “muddy the waters” ; the tool should be cutting corners and muddying waters and it can't becuse there's no tool that cuts corners and muddys waters and a non-existent tool cannot be a metaphor for anything. Except maybe an AI instructor or a Harvard education.
Using ‘s instead of is in important places in a sentence. I don't think Althouse does that.
the tool’s
AI’s
what’s
Sliding between two points
Point 1
tech students get scolded for using is now in the hands of their profs
some students smell hypocrisy
unease that the rules feel uneven
Point 2
[some students] fear a slide into impersonal learning.
students worrying about “authenticity.”
the human touch in teaching
Uncertain referent
"The real issue ... Is it the tech itself, or the signal it sends when profs lean on it?”
This sentence seems to be saying:
"Is the real issue the tech itself or is the real issue the signal the tech sends when profs lean on it." It sounds as if the "prof" is intermittently leaning on a fire alarm and the real issue is: it it the fault of the alarm or the prof that class keeps suddenly ending.
Also, Althouse doesn't present two sides of an issue and urge us to debate each other on the chosen significant topic. This is a mind that conceived the terms: "civility bullshit" and "cruel neutrality" and the sunrise run. And those rats. Grok never knew about the rats but if (preferred pronoun) is going to replace Althouse, (preferred pronoun) has to imagine the rats without seeing them. After all, that's what Althouse did.
Tina Trent: ten years ago I decided that the consensus practices of "minimal marking" and "let the students figure it out themselves" in teaching Writing were just excuses for professors to spend as little time on student feedback as they possibly could. So I experimented by going 180-degrees in the opposite direction: maximally marking-up, even copy-editing student papers and then "bribing" the students (with a small bonus on the grade) to implement my suggestions and edits. It is hugely labor-intensive, but the practice seems actually to work, as my students' writing improved significantly from assignment to assignment.
And it makes sense: when we teach music or athletics (two things we're actually decent at teaching), we tell students what corrections to make, and we keep repeating those corrections until they catch on. Why wouldn't the same thing work for Writing?
You won't be surprised to learn, though, that as the approach is very labor-intensive (I have to spend 10-20 minutes per paper) absolutely no one in Composition Studies was in the slightest bit interested in seeing if my results could be validated and my method developed further.
(n.b., this is NOT the 1960s and earlier approach of "a paper should fail if it has 2 comma-splices," i.e., just expect 'perfection' and that will show up. Students can and do make a ton of mistakes, but I tell them what the mistake is, why it's wrong, and how to fix it. Hopefully, then, they make fewer mistakes in the next paper).
Is Grok gender neutered?
Automated Intelligence (AI) is neither discerning nor creative. Anthropogenic Intelligence (AI) is two hops ahead, but a literary tortoise, which upon reflection enables the carbon bag of mostly water to win against the electronic hare in an intellectual race.
It's all about one's prompts. IMO Grok outperforms Claude, which I also use for editing.
Politicratus: “ interesting, even charming, translations of the family photo I chose as a Disney Cartoon, Warner Bros, Frank Frazetta”
Frazetta? Did all your kids wind up with gigantic muscles and bloody axes?
JSM
I thought Grok would know who was asking questions.
Prof. M. Drout:
The professors at Creighton always corrected for spelling, punctuation and grammar. I well remember a test or paper for theology class taught by a Jesuit. It was all marked in red where I got it’s and its wrong. Never made that mistake again.
Me thinks Meade is going to have to perform an intervention to prevent Althouse from self-destructing on the Grok thing.
I agree, Prof. Drout, which is why I stopped being an adjunct. At my last college, they told me to just show videos and not assign any writing. And I couldn't do that, so it made more sense to do something constructive like work at Taco Bell, where the pay was much better.
Post a Comment
Please use the comments forum to respond to the post. Don't fight with each other. Be substantive... or interesting... or funny. Comments should go up immediately... unless you're commenting on a post older than 2 days. Then you have to wait for us to moderate you through. It's also possible to get shunted into spam by the machine. We try to keep an eye on that and release the miscaught good stuff. We do delete some comments, but not for viewpoint... for bad faith.