Is It Even Possible To Ban A.I. In The Classroom?

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Mike Rugnetta (00:10):
Friends, hello, and welcome to Never Post, a
podcast for and about theInternet. I'm your host, Mike
Rugnetta. This intro was writtenon Tuesday, 09/23/2025 at
11:56AM eastern, and we have aslightly late show for you this
week. Sorry about that. I got atoddler cold, and we're in the
middle of renegotiating AIMike's contract, but we're here.

(00:32):
We made it. We did it. In ourthird ever show length segment,
contributing producer ToriDominguez Peak returns to look
at artificial intelligence inthe classroom, including one
instructor who has said no moreand has banned its use entirely.
Tori tells Jason what thatentails and tackles the old

(00:54):
Kennard is writing, thinking,and also bop spotter. But right
now, we're gonna take a quickbreak.
You're gonna listen to some adsunless you're on the member
feed. And when we return, we'regonna talk about a few of the
things that have happened sincethe last time you heard from us.
Hello? Is it five stories foryou this week you're looking

(01:16):
for? YouTube is reinstating thechannels of creators previously
suspended for violating COVIDnineteen and election
disinformation guidelines analphabet lawyer says are no
longer in force.
This according to a documentobtained by Fox News and
prepared for the US housejudiciary committee. Reflecting
on the company's commitment tofree expression, Daniel f

(01:38):
Donovan, counsel for alphabetrights, YouTube will provide an
opportunity for all creators torejoin the platform if the
company terminated theirchannels for repeated violations
of COVID nineteen and electionsintegrity policies that are no
longer in effect. YouTube takesseriously the importance of
protecting free expression, thedocument states elsewhere, and

(01:59):
access to a range of viewpoints.The document also explicitly
points out that YouTube has not,does not, and will not employ
any kind of fact checking orlabeling mechanism in its
software. At time of writing, nolist of the channels to be
potentially reinstated has beenpublished, but I bet it's not
that hard to figure out whomight be on it.
In completely unrelated news,Alex Jones recently appeared on

(02:24):
his show sporting a Hitlermustache about which he said, I
could tell you it had a wildeffect on women.

Jason Oberholtzer (02:31):
Ew.

The US Secret Service shut down a high powered
cellular network that theyclaimed posed a threat to tri
state area mobile communicationsthis week. CBS News reports
that, quote, law enforcementdiscovered 300 SIM servers over
a 100,000 SIM cards, enablingencrypted anonymous
communication and capable ofsending 30,000,000 text messages

(02:53):
per minute that could have,again, allegedly disabled cell
phone towers and launched adistributed denial of service
attack with the ability to blockemergency communications like
EMS and police dispatch, endquote. Secret service claims the
operation was well funded andpossibly under control of state
actors looking to cause troublefor UN week in New York.
Independent tech auditors andsecurity analysts are not so

(03:16):
convinced. Well funded, yes, butcapable of causing such
widespread havoc in New YorkCity of all places, not so much.
There is nothing about thisinfrastructure that would be
hugely disruptive or damaging tomobile phone networks, writes t
profit, the self describedtelecom informer for Hacker
Magazine 2,600 on Blue Sky.BookTok has managed to shoot

(03:41):
Timothy Snyder's lean butweighty 2017 book on tyranny to
the top of indie bookshop saleslists over the last few months.
On Tyranny's bullet point styleformat and short chapters,
writes Laura Miller for Slate,make it easy to break into
nuggets of exhortation. Aparticular favorite is lesson
number one, do not obey inadvance, urging individuals and

(04:03):
institutions not to appeaseauthoritarian governments before
they are even asked to. Somefans on TikTok temporarily turn
over their accounts to OnTyranny, reading one chapter
aloud per video until they'venarrated the whole thing.
Speaking of TikTok, an allegedso called framework deal has
been penned regarding the saleof the Chinese owned platform to

(04:25):
domestic concerns. The USgovernment and ByteDance have
brokered a forthcoming dealwhereby Oracle, Silver Lake
Technology Management, andAndreessen Horowitz would
oversee TikTok's US operations.This group would have an 80%
ownership share, and a member ofthe board would be appointed by
the US government. PresidentTrump has also suggested Fox
News Baron Rupert Murdoch willlikely be involved somehow. The

(04:49):
US based owners would leaseTikTok's infamous algorithm,
which Oracle would oversee and,quote, retrain.
Larry Ellison, CTO and founderof Oracle, has also recently
financed a number of large scalemedia mergers with his son,
David Ellison. ParamountSkydance controls CBS, Paramount
Pictures, and the streamerParamount Plus. The Ellisons are

(05:11):
also allegedly eyeing a takeoverof Warner Brothers Discovery,
which owns, among other things,CNN. And finally, get ready for
a really good sentence. Youready?
Ready for this good sentence?Here we go. Limewire, relaunched

(05:34):
as an NFT marketplace, haspurchased the rights to the
infamous fire festival brand.The New York Times reports that
the purchase was made for245,000 US dollars in an eBay
auction. It is unclear whatLimewire fire will become.
The music downloader turned NFTpeddler is apparently aiming for

(05:54):
something that, quote, expandsbeyond the digital realm and
taps into real worldexperiences, community, and
surprise, a thing which no doubtaligns well with the fire
festival brand. Ew. In show newsthis week, if you ordered a t
shirt, they are being printednext week. Once the print is
done, they will head toNeverpost HQ where they will be

(06:16):
packed and shipped one by one byhand with love. We will also
have very few stock designsavailable at the end of that
process.
I'm gonna let you know in theshow news portion of future
episodes when and where you cansnag those if you missed out.
But when I say very few, Ireally mean it. We're gonna
have, like, fewer than 10 stockshirts. And finally, holy cow,

(06:40):
we are a signal podcast awardfinalist in the technology
category. That is fun.
Heck yeah. If you could pleasego vote for us, we would love
that. We'll put a link in theshow notes. We are up against
some really rad folks, includingclose all tabs, who you may
remember from our hentai segmentand kill switch of whom we're

(07:00):
just generally fans. But pleasego vote for us.
We will love you forever. Signalawards, technology category,
there's a link in the shownotes. Okay. That's the news I
have for you this week. In thisepisode, Tory talks with Jason
about AI in the classroom.
But first, BopSpotter is aproject by Riley Walls, and it's
described this way. Somewhere inthe Mission District of San

(07:25):
Francisco is a microphonepointed down at the street
below. It is using a Shazammanner. So in our interstitials

(07:48):
this week, Hans took it uponhimself to recreate what he
imagined to be the sonicenvironment at the time of some
spotting. So what you are aboutto hear are not field
recordings, but carefullycrafted audio collages

Midnight, 12AM. 03:30AM. So I'm sitting

(10:32):
here at my desk today, watchingthe leaves slowly change color
when an email comes in, Fromfriend of the show, Tori
Dominguez Peek, who submitted tous a year ago a piece you might
remember, wherein AI chatbotcompanies reached out to her
with the proposition of turningher deceased mother into an AI
chatbot. Well, Tori is back withanother piece. I asked, what's

(10:56):
it about?
No one would let me know. Toriwanted to tell me herself. So
please, welcome back toNeverpost, Tori.

Tori Dominguez-Peak (11:04):
Hey, Jason. Thanks so much for bringing me
on.

I'm excited to learn what I'm about to
learn.

So today, I have a tale for you about AI on
college campuses and BrazilianPortuguese and solving crimes.

All three of my biggest interests. Let's get
started.

Story in three acts, the Aspheric life.
So Jason, like you said, fall isin full swing.

Absolutely.

I have I have purchased pumpkin spice
lattes. The leaves arecrunching. Sure. Hans has been
cooking beans.

Hans has been cooking beans. I'm up to five or
six layers every time I leavethe house.

And with fall happening comes a new
semester on college campuseseverywhere across The US. And so
kind of with that comes therenewed conversation that people
have been having about AI ineducation, and is it cheating,
and like all of the things.Sure. So earlier this year, the

(12:10):
New Yorker ran this piece withthe title, everyone is cheating
through college. And the wholecrux of it was just talking
about like how commonplace it isfor students to use ChatGPT or
to use it to like help withassignments or even going as a
part of like, write my termpaper for me.

Mhmm.

Blurring the line between getting it to
help you and then like, what hasbecome plagiarism. And like, the
most stunning part about thatarticle to me that I still think
about is they did a very smallsurvey. It was like a thousand
college students. But 90% ofthem had said they had used
ChatGPT to help with homeworkassignments.

And you found this surprising?

Yeah. Mean, just the the number was wild. I
knew it would be over 50. When Isaw 90, I was like, oh, we're
cooked. Okay.
But I just couldn't stopthinking about like, what
happens when we are letting apiece of technology kind of do
the thinking for us or do thetalking for us

Sure.

On that scale. And so I decided to talk
to someone.

Megan Fritts (13:17):
When I'm working on a new paper, I'll be, you
know, typing up some section andrealize I don't know how to
phrase it. And that tells me,oh, okay, I need to go figure
out what I actually think here.Because if I can't write about
it, that indicates a lack ofunderstanding there. So I think
what we're missing when we stopwriting ourselves is the ability
to check ourselves formisunderstandings.

So that's a professor I interviewed. She's
professor Megan Fritz. Sheteaches philosophy at University
of Arkansas at Little Rock.

I find it really interesting here that
she's using misunderstandings asa framing. As if like we are
interrogating our own brain whenwe set down to write. Yeah. That
feels pretty right to me. Tori,can I ask you a question
quickly?

Sure.

Did you cheat in college?

No. I cheated in high school though.

Did you cheat in ways that you think
fundamentally changed yourunderstanding or inhibited your
understanding of what you weredoing? Or did they just help you
get a better grade?

Okay. Let me just lay out the one scenario
that I cheated and I could helpyou could help me here.

Mhmm.

So I was failing chemistry. And so one
thing I noticed was that kidswho took a long time to take the
test, it was like third period.And then you had fourth period
and then there was lunch. And soI was like, oh, if I just take
forever on this test, I canfinish it later and study for it
during fourth period.

Yeah.

And so I was just like, oh, it's just
taking me forever. I have aheadache. Like, I guess I have
to come back for lunch. Andlike, come back during lunch and
finish up this test. Bell rings,I go to fourth period.
It's like my study hall period.I'm like study I'm like going
over the stoichiometry formulas.I'm hitting the books. And then

(15:04):
I lunch happens. I go back tothe chem room.
I actually reanswered some stuffbecause I was like, I have it in
my brain fresh now.

Beautiful.

And I took it. Is that cheating? It kind of
is.

I well, yeah. I mean, by the letter of the
law, but it's gamesmanship iswhat I think it is. Like, you
still

the player.

Like Yeah. You still walked in there with
the requisite knowledge or theunderstanding of how to find the
knowledge, and you applied yourbrain to the problems at hand
and Yeah. Got a better grade tothem. Under the framework that
professor Fritz is setting outhere, that seems to be a
different kind of malfeasance inthe classroom. And one that I

(15:41):
perhaps look more fondly on.
I cheated constantly.

Okay.

Either buy more time because I had not
prepared myself sufficiently ontime, or route myself around
rote memorization which Iconsidered to be an impediment
to learning and not a benchmarkby which you measured learning.
And honestly, resented having toregurgitate things that one
could find in a book onto a pagelater. So I count neither of
those things as cheating. Butlike the thinking that I had to

(16:07):
do with that information stillhappened in my head and hit the
page.

That's the thing. Like, I was still
studying.

Yeah.

I just convinced my chem teacher that I
had a headache when I didn't.Perfect. Right? Yeah. But I was
not plugging formulas in thechattypety and being like, what
are the answers to thesequestions?
Which I think is kind ofdifferent.

Sure.

And professor Fritz wrote this
article for the Chronicle ofHigher Education, and I will
read you out loud an excerptbecause I think it says
something like really poignant.It says, we're not simply
frustrated by just trying topolice AI use or the labor of
having to write up students foracademic dishonesty or the way
that reading student work hasbecome a rather nihilistic task.

(16:51):
Our frustration is not merelythat we don't care what AI has
to say and therefore get boredgrading papers. It is that we
actively miss reading thethoughts of our human students.

That is so dispiriting. Wow. Famously easy
job gets easier.

It's kind of a bummer. And she is like
hitting something here. Like,when you write down something on
a page, you are transmuting yourthoughts onto a page. And when
you turn it in, your instructoris reading your thoughts. Like,
we know there is a relationshipbetween writing and thinking.

Right. Exactly. And I like that she's
extending it to like arelationship between people on
either side of that activity.You will have a relationship
with your thoughts to thewriting and the people reading
your writing have a relationshipto those thoughts and therefore
you. And I think that isprobably one of the great joys
of teaching is to be inrelationship with those people
via their thoughts.

Yeah. And I I kind of got to this sort of
thing like writing and thinkingand like how they're related.
And I asked her, hey, is writingbasically the same thing as
thinking or are they kind ofintertwined in some way? And she
was like, yeah, they definitelyare related. And that we write
it to document our thoughts, butwe also write to come up with

(18:08):
thoughts.
It's kind of this really uniquelike both and relationship.

Yeah. You definitely you write to find
when you have to stop writingbecause you don't know what's
there, and then you have to gothink about it.

Yeah. I mean, whenever I write a script
for Neverpost, I will write likea paragraph and then walk away,
and then come back like two dayslater. And then, you know, it's
just it's a slow process. Butit's because I'm in relation
with my brain and trying tofigure it out. And then I'm also
doing something that is alsovery thinking heavy, which is

(18:42):
I'm learning a second language.
I spoke more Spanish as a kidgrowing up in a Latin American
household. And then I lost it asit became a teenager. And then
I'm trying to get back into itas an adult. Mhmm. And so like,
when I speak Spanish in my adultlearner's Spanish class, I'm
like having a thought in Englishand then translating it in my

(19:04):
head and then saying it out loudto them in Spanish, right, to my
instructor.
Sure. And then she sayssomething back to me in Spanish,
and then I'm trying to translateit into English in my head. And
it's just this very likemechanical relationship, and
it's not easy. And I feel alittle bit like a baby alerting
to speak for the first time.

Yeah. Okay. So if I'm if I'm hearing this
right then, it's like you'resort of seeing the the the gap
between the thought that you'rehaving and your ability to
articulate it in this secondsnow. I guess, re seconds, third,
second, again, language.Probably especially because at
one point, it was not there.
And you're feeling this, like,this break in the chain between

(19:46):
you having a thought and beingable to articulate that. Yeah.
And that feels sort of similarto what you think is happening
with the insertion of these AItools into the way people are
writing these days.

Yeah. And professor Fritz kind of brought
this up that AI is kind of thismiddleman between thought and
language that's never been therebefore.

The difference between being a native speaker
of a language and being someonewho's learned a language, I
think is is the perfect exampleof what we are risking, when we
use generative AI for ourwriting and, speaking, that we
risk going from this kind ofnative speaker status to a a a
situation where we have to if wewant to have these skills at

(20:27):
all, we have to reteach it toourselves in a in a really
artificial way.

So is the concern there almost like what
happens in its absence? Or likeit's like what happens if you're
without your Spanish Englishdictionary as it were Yeah. That
you'd actually don't havecontrol over the language.

Yeah. It it feels like if we outsource for
example, you didn't do thereading for your college class,
but you have to write paperabout it and you're just like,
shitty, this paper about thisthing. You didn't engage with
the text. Yeah. You didn'ttransmit the text into writing.
And so like, you're probably noteven remember what that class

(21:03):
was about. Yeah. You're losingsome type of critical like brain
step that helps you metabolizeinformation. Does that make
sense? Like, I think writinghelps you metabolize
information.
And like, if I could just becandid, I think writing is kind
of mentally painful for me.

Oh, yeah. That's the whole thing about
writing.

Like It

hurts and it's bad.

It hurts and it's bad, and that's the
whole point. And I will writesomething and I'll walk away and
I come I will come back threedays later, And I'm just like,
oh, this psychologically hurts.But then you get into a groove
and then it feels good andyou've kind of metabolized your
thoughts and it comes out and itfeels great. And there's just
such an emotional experience inthat. And so when you're just

(21:45):
like, hey chat, GBT, write thispodcast episode, summarize the
notes from this interview I havewith the source, or write the
interview questions.

Yeah.

You're kind of losing some threads.

Yeah. You know, not to be the person who's
constantly defending cheatinghere. But when you brought up
writing the report in the bookyou have not read, that is
something I believe a It's lotof us have a classic move. You
have to try it at some point.And when you do that, you
marshal enough information aboutthe book, you skim it, you look
up some notes, you try to findas much as you can to walk in

(22:20):
there.
But the thing you're doing whenyou walk in there is like using
your brain. It is learning. Itis performing. It is something
that requires you to undergo aprocess that will help you be a
better thinker and communicatorin the future because you are
actually doing a task. And tome, what feels scary about this

(22:42):
is that it is removing themental load of doing the task.
Mhmm. Not so much cheating likethe information, which is like
the veneer around which we alldo the process of learning, but
it's removing the actual mentaltask, which is the point of
sitting down and being a part ofa university or a class or
whatever the case may be. So Ifeel like this has to feel

(23:02):
different for teachers. Like,they've walked into campuses
every fall for millennia andbeen like, alright, everyone's
cheating. How do I make surethat I know that they are smart
enough to continue down the roadafter they continue cheating?
Like, I I that's probablyunlikely that people believe
they have a complete fail proofmethod to stop all cheating
forever. Do you think thatprofessor Fritz or other
professors are feeling like thisis a different kind of stop

(23:25):
cheating move they need to make?

Yeah. So professor Fritz has kind of gone
nuclear. Okay. She instated apolicy in her classroom that is
just like, I am banning all AIfrom my classroom. Okay.
It includes ChatGPT. It evenincludes like Grammarly, which I
use Grammarly to like make surethat my emails aren't misspelled

(23:50):
or whatever.

Oh, interesting. Not even

too many exclamation points. She's like,
nope, not even that. BecauseGrammarly can suggest rewrites.

Yeah.

And that in itself is kind of generative AI.

Yeah.

And so I asked her like, okay, so how are
you enforcing this?

Sure.

It just seems like a lot of work to
enforce this. And she has a veryinteresting way of going about
this. So at some point earlierin the semester, she has them
write these short essay typeassignments in class. They are
handwritten, and they turn itinto her right there, hard copy.
And so it's zero chance of AIuse.

(24:27):
She had you write with a pen andpaper. Yeah. And she keeps these
essays as kind of evidence oflike, this is how these people
write. Oh. This is what yourvoice, your narrative voice is
like.
And so then later on in thesemester, you turn in something
electronically and it's got AIwritten stuff, she's gonna be
like, this doesn't sound likeyou.

Okay. Does she do the comparison process
herself or does she let AI dothe comparisons?

So she kinda does both. Interesting.
She does use eight different AIdetection programs that she runs
it through, which is Wow. It's alot.

Does the school pay for these?

I don't know. That's great question.
Don't seem cheap, do they?

No. I would imagine not.

And so that she runs it through all of
those. She also just looks at itherself and is like, yeah, I
could tell.

Yeah. Of course. Right? Like, teachers
have been seeing this, like,forever.

But I did ask, like, can you just tell off
the bat of a student's writingas AI generated just by looking
at their paper? And she said,yes. And that there's usually a
couple of clues.

Okay. Is this where you're coming from my Em
dashes? Yeah. Alright. I'lllisten to it at least.

The first clue is she calls them 50¢
words. Weird, like formal wordsthat the typical 18 year old
undergrad would just not beusing. Okay.

An example of that is like the word want, w o n t,
where so you might use it in asentence like, I want to take a
walk in the morning. So it'slike talking like an
inclination. That's a 50¢ wordthat I I I would say most of my
students probably aren't justcasually using in their
reflection writings.

Now you're coming from my vocabulary.
First, my em dashes.

Are you want to use wants?

I mean, of course, I am. But like, where in
an academic setting would I everrun across the need to express
my feelings through the wordwant?

Yeah. Exactly. Clue number two, and
you were pretty right aboutthis, was Em dashes. Yeah. It's
not just any Em dash because I II feel you, like, it sucks that
em dashes have become like thisweird red flag and it's like, I
like a good em dash.

Sure.

She said that there's a type of em dash
that is a red flag to her.

An acquaintance of mine, another philosophy
professor, he calls thoseepiphany dashes, where you go
from an ordinary, you know,thought like it's not just a
walk em dash, it's abrainstorming session. This is,
you know, you're having anepiphany.

Interesting. Yeah. So there's like an emotive
component to the Emdash whenit's used this way.

Some some sense it sounds like a LinkedIn
post. You know what I mean?Like, has that type type of
cadence for it.

Right. It's like It's sort of a it's like
copywriting usage not Yeah.Yeah. Yeah. Oh, that's super
interesting.

And then obviously, the third main clue
is like she runs it through oneof her software programs and
it's like, bing, AI generated.But yeah, mean, it's kind of an
intensive process.

Yeah. It sounds exhausting. I guess for
everyone, I suppose. Like, arethe students having a good time
while this is happening?

They don't seem to be big fans.

Sure.

As you can imagine, a lot of them have
reacted by saying, well, I don'tget why you've banned it because
my other professors and otherclasses don't care. So like, why
should it matter?

Oh, jeez. Alright. Well, at least they're
still learning something andthat something is emotional
manipulation.

Yeah. I guess they gotta learn to read
the syllabus.

Yeah.

But it does kind of bring up something which
is that, okay, some professorsdon't care. Professor Fritz very
much does. Yeah. And so collegestudents are kind of navigating
this landscape where it's likein the same semester, they might
have someone who's a realstickler about this stuff. And
they may also have a differentteacher who doesn't care.

(28:21):
And it seems like they have tonavigate all these individual AI
policies.

Or just not use AI? That's one way to
navigate all of them.

Or just not use it at all. But I asked Fritz
about this like, oh, do you talkwith other professors about
handling this? And she said thatshe actually sits on a couple of
AI related committees at heruniversity.

I think instructor uniformity and solidarity on
this issue is pretty importantfor our students. I thought it
was a good idea, I was excitedto to to try to make this
policy. But what ended uphappening is that people just
had such different views on AIuse in higher education that it

(29:06):
kind of just turned into debateevery every meeting, and we we
have not yet made any kind of apolicy.

Okay. Well, I suppose that's predictable. It's
academia. It's meetings. It'sconsensus.
That is not necessarily easy todo, but, like, you know, I'm not
gonna tip my hand on where Istand on this. I'm sure everyone
is in deep mystery here. Butlike, if it's working for
professor Fritz, like, just lether do her thing. It seems like
it works.

Yeah. I mean, she thinks that she feels
that her system works well forher. Yeah. But her way of doing
things relies on knowing whatthese students write like, and
knowing what they don't writelike, and they're writing as a
sort of fingerprint. But whathappens when you can't really
tell and you have to do somedetective work?

Mhmm.

So I found a detective.

Alright, listeners. This is the most
podcast break we will ever doafter these messages, a
detective.

Rui Sousa-Silva (30:26):
Even though we learn the same languages from
the same books and we learn wecan find the same words in the
same dictionaries, the way eachone of us uses language is
different. So we have a let'scall it a different style of
using language.

So this is doctor Rui Sosasilva. He's from
Portugal. His whole training andjob is as a forensic linguist.
Oh. So like, literally, hisfield is all about confirming
the identity of who wrote what.

Woah. That's a rad job.

I know. It's such a cool job. And I
literally didn't even know thisjob existed until I started
writing this episode. And I waslike, this is amazing. Wow.
And being able to identify thenuances of what someone sounds
like, that identity is called anidiolect.

So ideolect is your own way of speaking or
writing the language. So it's asif your DNA was related to the
way you use language.

I believe it.

So I mean, like, legally, criminally,
historically, a forensiclinguist is the person you call
to match the fingerprints ofsomeone's writing. Right?

Woah.

And so at the same time, he's also a
lecturer. He also teaches, whichmeans that he also has to deal
with the issue of students usingAI in his class. Really also has
students write in his class sitdown and write. I know you're
not using AI because I can lookat you writing. Right?
But when they do that, he'snoticing something different

(32:10):
than Professor Fritz.

What we see nowadays, people interact with
generative AI so much thatpeople are starting to write
like machines. With some of mystudents, I know that they are
sitting an exam and I know theywere the ones who wrote the text
and still when I read the textit sounds as if it was generated

(32:33):
by a machine. And that's becausewe accommodate with other people
and we accommodate in the sameway with the machines we
interact with. So we tend toaccommodate so much to the
machine that we learn so muchfrom the machine that we start
start writing like machines. Sothis is a challenge at the
moment.

So is he saying that because we're
ingesting so much writing thathas been made in this process
that we are starting toregurgitate Yeah. That Yeah.
That checks out.

That's wild though, isn't it? Yeah. Because
then it makes me think aboutprofessor Fritz's class and her
methods, like, what if peoplejust start writing like
ChatchyPT?

Oh, boy.

Then at some point, it's just gonna get
harder to tell.

Alright. You've made an un virtuous cycle
here. I see what has happened.

You see what has happened. Yeah. And so
because Rui is a forensiclinguist, like he is the person
you can tell who wrote what, Iwas like, can you please give me
an example of influencing theway of like how machines
influence the way people write.And he said that when you speak
to ChatGPT in Portuguese,because he lives in Portugal, it

(33:45):
will sometimes reply using aBrazilian Portuguese dialect,
and that has consequences.

Yeah. One example is the way when when
you're writing in English, youusually say, if you want to to
list a set of points, you'llsay, firstly, such and such.
Secondly, such and such. And inPortuguese, usually you wouldn't
use the literal pronunciation ofthe adverb. But people are now

(34:14):
doing that and that's becauseinterestingly Brazilian
Portuguese does that and becausewhen you look at language
variants, I mean in Portugalyou've got about 10,000,000
speakers, if you go to Brazilthere are 200,000,000, so for
generative AI engines they feedon languages.

(34:35):
So they they are more likely tofeed on Brazilian Portuguese

than There's just more Brazilian
Portuguese language data out Andso when Chateapiti speaks
Portuguese, it sounds Brazilian.

It's usually Brazilian Portuguese, even
though nowadays you can ask towrite in European Portuguese but
every now and then there is aword in Brazilian Portuguese for
example. So the fact that it wasbased on Brazilian Portuguese
and the fact that BrazilianPortuguese uses that literal
translation of firstly,secondly, thirdly, now people
are writing like that.

Oh, that's so interesting.

Even native speakers of European Portuguese
are writing like that at themoment.

Wow. As I'm hearing that, I'm just thinking
that that is not necessarily thecanary in the coal mine, but
there's probably some bettermetaphor for it just being the
visible object of something thatis also happening under the
layer of language and thinking.If we are regurgitating
different cultural grammarrules, we're probably also
surfacing other imports that wedon't know are imported from

(35:37):
different places.

Yeah. I mean, Brazilian adverbs are
pretty low stakes. I mean, Ithink it's kind of amusing. But
remember when I said thatforensic linguists identify the
DNA of language.

Yeah.

And they do that in criminal settings. Can I
outline a scary scenario foryou, Jason?

I'm braced. I'm ready.

Okay. So doctor Sosasilva told me that
one of the things that hiscolleagues talk about all the
time is the use of AI incriminal activity. So like using
generative AI to impersonatesomeone's writing style to write
something incriminating.

Sure.

So for example, somebody does not like
you. And so they write astrongly worded threat to a
politician, but they write itJason Operholzer style. And they
like, maybe they make a sockpuppet social media account and
they impersonate you, and theypost it on there. And you get a
visit from the police, andthey're like, you wrote this.

Yeah.

How do you prove that you didn't?

That's a good question. I would probably at
this point, try to point to likethe corpus of available writing
that I would have on hand. Like,I'm trying to really take a
cynical view of this and justassume that the the
infrastructure is weightedagainst me on this one, and
there's now a letter out theresigned by me that says, hey,

(37:02):
buddy, I'm gonna kill you. Andto disprove this, I don't think
I would have successful timeattacking the language on a word
by word basis. I would probablyhave to compile my own corpus of
writing, and I'd probably haveto divulge repositories of data
that I would otherwise want tokeep secret, like private

(37:23):
messages and be like, I willtake all of my eye messages and
put them in a model.
And you can see the way that Icommunicate and you can see like
all the communication I havehad, all the cynicism I have had
around politicians and thegovernment. And you tell me
where in this trajectory isthere the leap to a murderer.
And it is less about word choiceand more about state of mental

(37:47):
well-being. Like, is this theand then I'm going to write a
politician and murder themtrajectory? And here's every
piece of written correspondenceI have available to you.
And just hope that I've got adoctor like the good Portuguese
doctor doctor Sosa Silva on myside who can help me make a
better argument about thatmaterial than whoever's on the
other side.

Yeah. I mean, definitely, you would like
to call doctor Sosa Silva.Right?

Yeah.

They can analyze the text. They're like,
this is this isn't quite reek ofJason. There's something just a
little bit off of this.

Speaking of word choice, can we do better
than reek of Jason?

Reek of Jason.

For want of a better term, I suppose you can
keep reek.

So, yeah, this is where forensic linguist
come in and then you don't getarrested, hopefully. Right?
Okay. But as generative AI getsmore sophisticated, Rui thinks
that his work is going to getharder.

The developments in generative AI
will make it more complicatedfor forensic linguists to
attribute texts, which in turnwill mean that forensic
linguists will need to do moreresearch and to further their
research and to have more finegrained methods of attributing
authorship. But there willalways be a distinction between

(39:04):
the way humans produce text andthe way machines generate texts.
So things generative AI willevolve, forensic linguistics
will evolve, but eventually wewill always be able to pinpoint
differences between the texts.

So very similarly to the classroom here,
this seems like it is justcreating piles of work for
everybody.

Yeah. It seems like things are just gonna
get harder for everyone, whichis kind of a bummer. And I know,
like, the threatening thepolitician, like, that's a very
dramatic example, obviously. Butlike you said, with the
classroom, the idea I keepcoming back to is cognitive
offloading and like, not doingall these mental processes and

(39:50):
offloading it to AI, which thenI guess makes it harder for
forensically. Like, there's beenstudies from Microsoft, from the
SBS, Swiss Business School abouthow people who use generative AI
regularly tend to score lower onmarkers of critical thinking.
Like, there's actual data wehave now. Mhmm. Again, I'm

(40:12):
painting kind of a scary pictureto you. It's like, okay, so our
critical thinking may be gettingcompromised by a technology that
is also getting moresophisticated at pretending to
be us. And we're also startingto become influenced by the way
it writes.
Mhmm. Like, that's just such aweird trifecta.

Right. But doesn't the cycle also work in
the other direction? Like, weare the corpus of information
that the generative models needto continue their work. And as
we lose cognitive functionbecause of offloading, the
material that we are able tofeed depreciates in value as
well, which one imagines leadsto worse outputs from machines,

(40:55):
which we are synthesizing andfurther inhibits our ability to
think and provide a reasonablecorpus of information updated to
the moment from which Yeah. Themodels can select.

It's kind of like this feedback loop.
Yeah. Us feeding it and then itinfluencing us, and then all of
a sudden, we're just allspeaking Brazilian Portuguese,
right?

Yeah.

And getting accused of crimes we didn't
commit. And professor Fritz'sconcern about the feedback loop
is that it affects everyonedifferently.

A lot of people defend, letting students use AI
in their work, by saying thatthey see it as a tool for
equity. Maybe for students who,had a less privileged primary
education or students for whomEnglish is a second language, I
would contend the exact oppositeis true. That what this is doing

(41:48):
is setting the stage for genuinereading and writing skills
becoming something that isreally only accessible to the
elite class, those with a lot ofmoney and leisure time to
cultivate them intentionally.And so that's really something
that concerns me quite a bit.

Okay. I'm beginning to see why there's
such difficulty formingconsensus around this. I mean,
initially, I'll be honest, myreaction was yeah. It's obvious
that in a space where you'resupposed to be practicing
thinking and metastasizing yourown thoughts that AI is just not
helpful. It's not there for anyreason except for you to finish
the paper, which is arepresentation of the thoughts

(42:26):
you were supposed to be having,and it is like the wrong
takeaway from what you're doingin the classroom.
But now that they're applyingthese frameworks around
accessibility, I can see itbecoming a little more
complicated. I guess a lot of itboils down to how much you think
the role of the academy is toprepare you for work,
employment, and how much it isto help you engage with how you

(42:49):
think and learn. But I'mbeginning to see why the stakes
are a little more complicatedthan perhaps they feel at first
blush.

Yeah. And then when you think about
professor Fritz's role in thecommittees and how it's just
really hard to come to agreementor make any sort of policy, I
think this is something thatthey're gonna continue to
wrestle with for a long time.

As for myself, my policy won't be changing. And
that's, you know, that's aboutall I can do about that.

Tori, thank you as always for bringing in a
really insightful piece here.

Yeah. Thanks so much for letting me
talk about it.

Yeah.

And I just wanted to shout out thanks to
professor Fritz and to doctorSosa Silva.

At least one of whom is probably going to be
getting me out of jail over thenext couple years. So pre thank
you for that one.

Pre thank you for that one.

Tori, where can folks find you and all of
the writing that definitivelycomes from your own brain on the
web?

You can find me at toori d p nine
8.bluesky.social. And you canalso find my podcast about video
games that I make of my ownbrain and play with my own brain
atpress-startpod.bluesky.social.

Tuesday, September 23, 10:30AM.

That is the show we have for you this week. We're
gonna be back here in the mainfeed on Wednesday, October 8. We
are proud and thankful andextremely lucky to have the
member community that we do.Without the support of our
members, this show would not andcould not exist. So I just wanna

(47:56):
say thank you.
If you would like to become aNeverPost member and join this
community for as little as $4 amonth, you can do that at
neverpo.st. Where also if amembership is a little too big
of a commitment in these strangeand trying times, you can also
tip us a one time any dollaramount, and we promise that we

(48:20):
will spend every last red centon arcade games chewing gum and
baseball cards. Become a memberat neverpo.st. Never Post's
producers are Audrey Evans,Georgia Hampton, and the

(48:41):
mysterious, doctor first name,last name. Our senior producer
is Hans Buto.
Our executive producer is JasonOberholzer, and the show's host,
that's me, is Mike Rugnetta. Andthen this warning flashes on the
light meter. Inside the house, apilot light is always burning in
the oven's eyes. The low roof ispulled down over the eyes like a

(49:04):
hat. And underneath thewarnings, light motif networks
of subterranean lines run likethe nervous system or bloodlines
or fractures spreading fromtectonic lines of fault.
In distant coasts, heavy andlight petroleum is piped across
state lines and gas, electric,oil, and water lines convey

(49:25):
their vital humors to the house.Excerpt of nervous systems by
Greg Williamson. Never Post is aproduction of charts and leisure
and is distributed byRadiotopia.

All Episodes

Episode Transcript

Popular Podcasts

Stuff You Should Know

The Joe Rogan Experience

On Purpose with Jay Shetty

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Is It Even Possible To Ban A.I. In The Classroom?

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}Stuff You Should Know

The Joe Rogan Experience

On Purpose with Jay Shetty

All Episodes

Is It Even Possible To Ban A.I. In The Classroom?

Stuff You Should Know