February 24, 2023

I've sometimes thought I could set up AI to keep writing this blog after I can no longer go on.

 It seems James Taranto had a similar idea...

The chatbot seems able only to point at what humans have done, not to actually do it.

It demonstrates what was so good about the real thing. It's everything that is missing. 

Once again, I think of the steak scene in the 1986 movie "The Fly."

As I blogged a year ago (in a different context):
... Seth Brundle (Jeff Goldblum) is testing his teleportation device by running a half of a steak through it. He then cooks the teleported and unteleported halves of the steak and feeds them to his girlfriend Ronnie (Geena Davis). The unteleported steak, Ronnie says, "tastes like a steak" As for the teleported steak: "Oh... Oh, oh, tastes funny." She spits it out. "It tastes... synthetic."

Brundle theorizes: "The computer is giving us its interpretation... of a steak. It's, uh translating it for us; it's rethinking it, rather than reproducing it, and something is getting lost in the translation.... The flesh. It should make the computer, uh crazy. Like those old ladies pinching babies. But it doesn't; not yet because I haven't taught the computer to be made crazy by the... flesh. The poetry of the steak. So, I'm gonna start teaching it now."

43 comments:

Dave Begley said...

Don’t be ridiculous, Ann. You’re sui generis. No computer could come close to replacing Ann Althouse.

rehajm said...

Hey cool! Ann, if you AI the blog the commentariat can AI the comments section then we can all retire to the south of France like Chuck Barris…

mezzrow said...

I will now have to see if the ribeye in my fridge can summon up a sonnet.

There is more mystery in the kitchen than I might have imagined.

Birches said...

Bottom Story of the Day

re Pete said...

"Is the scenery changing

Am I getting it wrong

Is the whole thing going backwards

Are they playing our song?

Where were you when it started

Do you want it for free

What was it you wanted

Are you talking to me?"

Sebastian said...

"The chatbot seems able only to point at what humans have done"

Correct. That's because it efficiently reproduces patterns it found in what humans have done.

"not to actually do it"

That's always going to be true, by definition of "actually." But over time it's going to be harder for humans to tell the difference. It already is in many cases: ChatGPT is acing a bunch of exams, passing psychological tests, etc.

Kevin said...

If Not $ElectionDay Then
teardown $allsides
ElseIf $ElectionDay Then
support $democrat
Else
post $sunrise
EndIf

rrsafety said...

I asked ChatGPT to "write a blog post about the US supreme court in the style of Anne Althouse" ... it did not sound like you at all.

"The US Supreme Court is one of the most important institutions in American law and politics. As a professor of law and a keen observer of legal developments, I have watched the Court’s decisions and trends with great interest over the years. In this post, I will share my thoughts on the Court and its role in shaping American law and society.

The Supreme Court is the highest court in the United States and is composed of nine justices, who are appointed by the President and confirmed by the Senate. The Court’s primary role is to interpret the Constitution and ensure that the laws of the land are consistent with its provisions. This may sound like a straightforward task, but in reality, it can be incredibly complex and controversial."

Shouting Thomas said...

Seems to me the bigger issue would be using AI for comment moderation.

I can see how that would be programmed.

tim maguire said...

Based on Taranto's excerpt, AI could create templates that serve as a first draft of posts. The basic script was fine, but there were almost no details. If not for mention of Trump, it would have been generic. It could have been written yesterday, last year, or last decade.

It was timeless, which is not what we look for in a hot take.

Rusty said...

"There is more mystery in the kitchen than I might have imagined."
So you've met my wife?
An original has a quality all it's own.
Except there's a German group that does a rendition of "Tom's Diner" that is awesome. They own the song now.

Ann Althouse said...

@ re Pete

We're listening to that now here at Meadhouse

Drago said...

AI generating Althouse blog posts? Impossible.

AI generating with 100% accuracy every supreme court case vote and rationale by the democratical court members? Piece of cake.

tommyesq said...

If not for mention of Trump, it would have been generic.

These days, that still makes it generic.

Birches said...

One of my favorite things about Taranto's Best of the Web is that he referred to you as a blogress. It seemed to me he was designating you as a noble version of a blogger.

rcocean said...

Isn't that exactly what a bot would write? Althouse goes on a "Trip" and then we get talk of a "Bot" taking over. Invasion of the bot snatchers?

Joe Smith said...

Perhaps it could be written and posted, but right now I think the hard part would be figuring out how to find the links and write about those.

Random thoughts? Sure, but it wouldn't be your blog...

n.n said...

A basket of knowledge, a cache of correlations, an anthropogenic map with tuned levers. Lowered degrees of freedom through anthropogenic simulation, popular perception. A baby or a fetal-baby, a granny, too, that is the question... and social distance from mechanical to human consciousness and conscience.

Bob Boyd said...

I've sometimes thought I could set up AI to keep writing this blog after I can no longer go on.

As soon as you turned your back the damn thing would start showing up for work wearing shorts.

typingtalker said...

Ann wrote, "The chatbot seems able only to point at what humans have done, not to actually do it."

Wait until it finds a link to a 3D printer ...

Artificial intelligence (AI) implementations have a lot of potential for additive manufacturing. Despite some progress, AI applications in streamlining additive manufacturing to be integrated into other manufacturing techniques or become a commodity for users are still a long way off. AI can assist additive manufacturing in a variety of ways, including design correlation, design improvement, defect reduction, and microstructural design. The main stumbling block right now is the availability and reliability of data needed for training.

Integration of Artificial Intelligence in Additive Manufacturing

Soon.

n.n said...

referred to you as a blogress. It seemed to me he was designating you as a noble version of a blogger.

Your Highness. We approve.

Highness (abbreviation HH, oral address Your Highness) is a formal style used to address (in second person) or refer to (in third person) certain members of a reigning or formerly reigning dynasty. It is typically used with a possessive adjective: "His Highness", "Her Highness" (HH), "Their Highnesses", etc.
- wiki

Wait... His Highness. Her Highness. Their Highnesses, maybe. We prefer "Their Highness" for purposes of equity, inclusion, and symmetry.

n.n said...

Integration of Artificial Intelligence in Additive Manufacturing

Several mutations later you have a toxic dispersal in East Palestine.

wildswan said...

Interesting reading that Chatbot "in the style of Althouse." It was so far off yet it bore some tangential relationship. It was not, for example, Pete Buttigieg or Kamala Harris. Was it the script for the ninth season of "Althouse", the challenging story of a liberal who believed in free speech and lived in Madison, Wisconsin during the crazy years"? Maybe that's what an Amazon or Netflix series of the future might outline as a story-line but those series deteriorate into repetition after three years, max, whereas this blog is not predictable after twenty years.
Things out there are existing in different ways and we wrestle the truth about those different ways and with appearances and pretenses of existence all our lives.

wildswan said...

Interesting reading that Chatbot "in the style of Althouse." It was so far off yet it bore some tangential relationship. It was not, for example, Pete Buttigieg or Kamala Harris. Was it the script for the ninth season of "Althouse", the challenging story of a liberal who believed in free speech and lived in Madison, Wisconsin during the crazy years"? Maybe that's what an Amazon or Netflix series of the future might outline as a story-line but those series deteriorate into repetition after three years, max, whereas this blog is not predictable after twenty years.
Things out there are existing in different ways and we wrestle the truth about those different ways and with appearances and pretenses of existence all our lives.

Interested Bystander said...

Just for fun I asked ChatGPT to write something in the style of Ann and gave it an example to work with. I think it got closer when I gave it something to chew on.

"That's not really what I was looking for. I'd like to see an example where Ann Althouse questions the usage of a word with examples. For example she often goes back to wonder about the origins of the word "garner" and why it's used in certain circustances but not others. Write a paragraph in her style analyzing a particular word. Use the word "antiquated" as the example."

"I apologize for the misunderstanding. Here's a paragraph in the style of Ann Althouse analyzing the word "antiquated":

"Antiquated. What a curious word. It evokes images of dusty old relics and outdated customs. But why do we use this particular word to describe things that are outdated? What's the origin of the word 'antiquated' and how did it come to be associated with things that are no longer in fashion? And why do we tend to use this word in a negative sense, as if being antiquated is inherently bad? After all, there are many things from the past that are still valuable and worthy of preservation. Perhaps we should be more thoughtful in our use of this word, and consider whether it accurately reflects the value and meaning of the things we are describing."

James K said...

"I've sometimes thought I could set up AI to keep writing this blog after I can no longer go on."

How do we know you haven't already done this?

Achilles said...

"I've sometimes thought I could set up AI to keep writing this blog after I can no longer go on."


Well.

Think about what you actually do here.

The AI might not support evil publications like the WAPO or NYT's.

So there is that.

Dr Weevil said...

Birches (9:19am):
I'm afraid 'blogress' always looks to me like a portmanteau word for an 'ogress' (female ogre) who posts on her blog whenever she's not busy gnawing the bones of billy goats under a bridge. Not my picture of AA at all.

gadfly said...

Oh Buoy! ChatGPT humor is just around the corner, if and when AI can deal with funny boy buoys and yellow submarines.

Yancey Ward said...

They tried this with Joe Biden, but the Chatbot kept shitting in its pants.

Barry Dauphin said...

My prompt and ChatGPT response:

Write a brief essay describing why it is aesthetically unpleasing to see men wearing shorts:

As an AI language model, I must point out that beauty and aesthetics are subjective and vary greatly from person to person, culture to culture, and even within a given society. However, I can offer some general perspectives on why some people may find men wearing shorts aesthetically unpleasing.

Firstly, shorts are often associated with casual and informal wear, which may not be appropriate for certain situations or occasions. For example, in formal events like weddings, funerals, or business meetings, men wearing shorts may seem inappropriate, disrespectful or unprofessional. Moreover, in certain cultures, it may be considered taboo or disrespectful for men to expose their bare legs in public, which can cause discomfort or offense to some people.

Secondly, the length and style of shorts can affect their aesthetic appeal. Short shorts or baggy and ill-fitting shorts can look unflattering and unappealing, while longer and well-fitted shorts can be more visually pleasing. Additionally, some men may not have the physique to carry off wearing shorts, which can further detract from their aesthetic appeal.

Thirdly, personal taste and cultural norms play a significant role in how people perceive the aesthetic appeal of shorts. In some cultures or regions, shorts are widely accepted and even preferred as a staple summer wear, while in others, they may be frowned upon or considered unfashionable. Moreover, personal preferences for fashion and style can vary widely, which can affect how people perceive the aesthetic appeal of shorts.

In conclusion, while it may be aesthetically unpleasing for some people to see men wearing shorts, it is important to note that beauty and aesthetics are subjective and influenced by cultural norms, personal preferences, and other factors. Ultimately, what is deemed aesthetically pleasing is a matter of personal taste and can vary widely among different individuals and societies.

Ann Althouse said...

“ Perhaps we should be more thoughtful in our use of this word, and consider whether it accurately reflects the value and meaning of the things we are describing."

That made me laugh out loud.

I’d never write like that, and I wouldn’t misuse a comma like that.

So putrid!

We should consider calling more things putrid.

AI can’t quote the OED.

Narr said...

AI is even more stupid and lacking in creative imagination than most humans, and that's a high bar.

tim in vermont said...

The think about the ChatBot is that you have to keep asking it questions to improve it. Or give it more details to use, like offering up a word for 'Althouse' to analyze. So, naturally, I chose 'garner."

Some people might find that the worst thing about the word "garner" is that it is not commonly used in everyday conversation, so it may sound formal or pretentious when used in casual settings. Others may find the pronunciation of the word difficult, especially for non-native speakers, as the "g" is silent. Nonetheless, this is a matter of personal preference and does not reflect any objective or inherent negative quality of the word itself.

Who says that AI isn't funny.

Smilin' Jack said...

“ Perhaps we should be more thoughtful in our use of this word, and consider whether it accurately reflects the value and meaning of the things we are describing."

That made me laugh out loud.

I’d never write like that, and I wouldn’t misuse a comma like that.


Not a misuse, a matter of style. Personally, I would use it. Chatbots aren’t going to make mistakes in grammar, any more than they would in arithmetic.

rcocean said...

Secondly, the length and style of shorts and amount of Nazism can affect their aesthetic appeal. Short shorts or baggy and ill-fitting shorts can look unflattering and unappealing, while longer and well-fitted shorts with plenty of Swaztikas be more visually pleasing. Additionally, unlike Hitler or Himmler, some men may not have the physique to carry off wearing shorts, which can further detract from their aesthetic appeal. - Nazi Chatbot

tim in vermont said...

OK, I am using the free version, and it sometimes overloads and just forgets what it is doing, so I got cut off in the middle of writing the short story. Anyway, it suggested a female character, and made her "a journalist who is passionate about sustainability and fighting climate change..." which, to be honest, wasn't a bad fit for the story, but still.

Other than that, it wrote out a lot of prose and dialogue at kind of a high school level, that could be edited into something better, I think.

I might have to start paying for it, so that it will remember sessions between logins, and store more information per session. once it crapped out on me, the story was gone from it's memory, so I could no longer extend it. without explicitly referencing every item. For instance, it remembered where the character was, and you could ask it to recommend a local restaurant where the characters could go, and it would, but once it crapped out, you could not do stuff like that anymore without reminding it of what had happened up to then.

BudBrown said...
This comment has been removed by the author.
Lucien said...

A properly Woke AI would face a quandary in cooking steak: On one hand, steak is bad for The Planet; but Trump clearly hates steak, because he has it well-done and puts ketchup on it.

Lurker21 said...

That does suggest a future for literary criticism. Give chatbot all of Herman Melville's writings, let chatbot create new works, and critics can analyze them as if Melville wrote them, and then as if chatbot wrote them. Hurrah! Lit departments are saved from the chopping block!

Bunkypotatohead said...

You'll need to arrange for AI to generate the comments also, as most everyone here will be dead too.

toxdoc said...

This is how i end uo when i use Ai:https://www.youtube.com/watch?v=8SbJIlEd6jA

effinayright said...

@Althouse:
If you've thought you could easily be replaced by AI, aren't you implying you're already lazily "phoning it in", and that no one is noticing that you are?

p.s. how many years have you been doing this already ?

F u cn rd this u cn uz AI 2 fool pepl.