Hey, what's going on, everybody?

On today's show we are talking about the ghost in the machine, how AI is Co authoring human language.

So part of that is to be talking about the lovely M dash, but not just the M dash, it's the one that gets a lot of love and hate right now.

But we're going to be talking about how that, like seemingly trivial punctuation mark, has become the unlikely cultural flashpoint in the age of AI.

The M dash was once, you know, a flourish of literary prose, and now it's viewed by many as a telltale sign of machine generated text.

Online forums and social media platforms are ripe with users claiming they can spot AI writing, but it's over reliance on this single character, leading to a phenomenon some of dubbed M dash fatigue.

This debate, however, and all of today's show is not merely A squabble over punctuation.

It's a Canary in the coal mine singling a much larger, much more profound shift, the emergence of the distinct AI infected, reflected infected, or reflected language that is seeping into our daily communication.

The M Dash controversy reveals our collective anxiety and curiosity about how AI is influencing our most human trait.

So we're going to be diving into all of that that today and much more.

This is the daily AI show.

Today is August 5th, 2025.

I'm currently joined by Jimmy and Andy and I'm Brian.

Welcome, guys.

I think we might have Carl Poppet in the door here maybe a little bit today, which would be great to see him.

The idea for this show today was actually from Sam Altman and not that he had specifically spoken about this larger conversation that I want to have about how AI is influencing human language and and all the things that go along with that.

There's many other different parts and pieces to it.

What he actually talked about just sort of as a kind of, I think it was on the Via Theo Von podcast and he sort of, I don't want to say flippantly, I don't know that's the right word, but he sort of like casually talked about the M dash and he's like, yeah, we got to fix that.

He's like, I think that's something we need to fix in the model.

But it was sort of like just a casual thing.

He's like, you know, people wanted it, so we had it in there and now people don't now it's too much.

And maybe I think we've gone too far the other way.

And so I walked away from just that thought and I went, I think there's something bigger here.

Yeah, it's easier to point at Delve, the M dash.

There's several others that we could probably point out that I know as far as my reach that came up quite a bit that we would say that's AI speak.

But I think the the larger conversation for me is, you know, probably much in the way the typewriter was perhaps rejected by some people who said, you know, this is flying in the face of, you know, great prose and handwritten letters and what is this mechanical machine going to do for the great literary writings of stuff?

And I am sure there was there was pushback against that.

I am sure the invention of the QWERTY keyboard, which was done because of jamming on the machine and not necessarily because it's super efficient, also probably changed the way people wrote.

So it's not as if this is unprecedented that we would somehow have some new tool that shapes or or molds the way we write or speak.

That happens all the time.

I don't speak the same way I did when I was a kid in New York.

I use different words where even though our moms told us we were special butterflies, we're, we're also parrots, right?

And so we were more likely to parrot back out what our surroundings are.

But what's different about AI?

What I'd love to like, kick off the conversation with both of y'all is it's happening so fast and it's becoming such a major part of the way that we write, produce reports that today isn't about like, should it or should you write full papers and essays in college or or otherwise and do professional reports of AI?

That's to me a foregone conclusion that AI will be used.

The question is, are we OK with the fact that AI is perhaps adjusting and shifting and moving us towards human language?

And the bigger question is a lot of this is trained on English and sort of the median English at that, which means that there's a whole lot of people who are left out of that conversation.

And so even though for me as a white American male, you know, it probably is not going to be as big of a deal, but for people who come from other cultures, other races, other religions, other languages, whatever the case may be, I would imagine it's a bigger impact on them.

And then you get into this idea of is it just mixing everybody's AI going to mix everybody into a soup?

Because if I had a typewriter and I was able and it would had it had letters on it, not like Chinese characters or something like that.

It was a QWERTY keyboard.

Well, I can write in French, I can write in Spanish, and I can reasonably use the same keyboard to do that.

This is different.

And is it bigger?

Is it bigger than just people being upset about the M dash or whatever?

The next thing is the M dash because it will inevitably be something else.

So that's sort of where my brain is on this.

I would like to give sort of a brain dump of like why I think this is sort of a big deal and what we need to discuss.

But what do you guys think?

You know, is it a concern of yours?

Is it something that we should be paying more attention to?

Or is this just like the next tool, like the typewriter, except that it's more ubiquitous, it's being used more now?

I'll jump in and say I'm, I'm, I'm, I'm first of all, I'm bringing my Dartmouth English major perspective onto this one.

And so I, I'm declaring authority of some sort here and one of the major authorities on style of writing that you if among us English major.

So it's a very familiar tome, Elements of Style by Strunk and White's required reading.

And you know that follows a a very simple directive that Strunk put forward, which is omit needless words.

Vigorous writing is concise.

A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts.

End Quote.

So that is really what the machine that AI is doing is it's not following that specific directive, but it's it's been trained to do something very specific.

It's supposed to be helpful, explanatory and accessible and conversational.

By accessible and conversational, we mean let mimic to the best degree possible in whatever language you're speaking, a natural and conversational tone while restricting your outputs to something that's explanatory.

So it's no surprise to me that it's using M dashes.

M dashes are more reflective of conversational rather than formal exposition in writing.

OK, so the M-I, I use the M-IN my writing and and I am, I'm offended that there's such a backlash against it.

And then I'm going to be accused of being an AI automaton because I use it.

Look, the dash is there to express a clear break in thought, and you can use commas.

That's the formal way.

If you were writing an academic paper, you would not use an M dash.

You would have to use commas, semicolons, and colons correctly.

But a dash is done for emphasis and it's done as an explanation, like a parenthetical explanation, but without de emphasizing as parentheses do like, oh, this is just an aside.

That's what a parens means.

But a dash says, no, this is mainline.

But I'm going to explain something to you here.

That's a part of this expression that I'm making here.

So anyway, why does, why does AI use M dashes?

And I'm setting aside for the moment that, you know, the, the AI can inadvertently, because of its focus, it can shift, you know, imperceptible shift our way of speaking.

And, and that could have some, some, you know, longer term consequences, But I don't think it's for the, the worse, It's for the better.

Now there is a need to maintain some stylistic variation and, and you know, authenticity and the way you speak and the way you write, but that's more for entertainment writing in my view, than scientific and explanatory AI purposes.

So anyway, why does it use M dashes?

I I asked this question of perplexity because I wouldn't think that most of the of the training data had a surplus of M dashes in it, but it hastened to correct me.

It said, look, novels and formal essays, if they were a part of the corpus, would use fewer dashes.

That's just not conversational.

That's a more formal form of writing.

But AI training like likely emphasized Wikipedia, which is definitions and clarifications, and that would use dashes effectively.

Educational content, which uses dashes for explanatory interjections, and technical documentation where specifications and examples are set apart by dashes.

And by the way, modern journalism uses them in order to adopt A more conversational and accessible tone.

So AI has a dash preference?

Great, it's good.

And by the way, when you start to use voice as the interaction mode, you'll never notice whether it's a dash or a comma.

So why does it matter?

Well, I'm glad you brought up the like the the use of the M-I actually read something really interesting about this that although it can seem like personal, it's a personal attack.

I OK, so I'll just say I'm the opposite of you, Andy.

I never used an M-I don't even think I'd heard of an M dash before it became an issue.

Kind of like delve.

I don't think I ever used the word delve until it all of a sudden it was everywhere and people are like, Oh my God, it won't.

I I used it.

I I I can't use it anymore because it makes me sound like an automaton.

And and Beth isn't here today, but Beth has said the same thing.

She's like, you know, over my cold.

And she didn't say this.

I'm paraphrasing, but it was like over my dead body.

You're taking away my M dash.

I use it and I like it and I'm even like, I don't care.

And this could be a a a flash in the pan and this may not last forever.

And it could it could certainly die down.

And I think the use of the M dash will will be more generally accepted as AI has moved on to the next thing.

That's a problem.

But what I wanted to bring up really quick before we Jimmy, we get your thoughts, are I read something really interesting that is, you know, the M-IS just as much mathematical and mechanical as it is something that we perceive as human.

Like we read what's coming out of the algorithm.

We go, oh, that sounds very human.

Maybe I wouldn't have said it that way, but it's giving me human.

And so it's very easy, obviously, to put human traits on the algorithm.

We do that all the time when we say things like think.

Well.

Right.

Like, is it thinking or is it just a mathematical really smart algorithm?

Well, I don't know.

It's probably the latter, right?

Well, M-IS are a way for the algorithm to connect through tokens.

And there was actually a lot of research done that I found really interesting about when even small words like to or a, which could potentially be stripped out and still be concise to your point, Andy, but it actually can negatively impact the algorithm's ability to reproduce and put out quality content.

So we look at the M dash and go, that's annoying.

Maybe if you're me and I go, that's annoying because I don't speak like that.

I do not write like that.

And so this is something that's new that I'm always having to strip out.

And now I have to strip out because, well, I don't want I don't want writing that I've co-authored to feel like it was 100% AI driven, even if I worked on it and I had wordsmithed it.

But that did that.

It's perhaps more than that, that that M-IS serving an actual function of connectivity inside the algorithms to ensure that you're actually getting higher quality.

So I found it just interesting to the idea that while it's annoying to me, what it might actually be doing is serving a higher purpose of giving me higher quality answers.

I'm annoyed that the M-IS in there, but to not have the M-IN there actually might have produced a lesser quality writing.

I just found that fascinating because like it annoys me, but at the same time, it might actually serve a good purpose that the overall quality of the writing is superior because it's there, because it's being used as a tool there.

I don't know.

I just, I was like, I would not have thought of that.

Jimmy, what's your thoughts on it?

Well, I mean, we, we know that because LLMS are the predictive model.

It's, you know, it's predicting the next token, what's going to be most used.

Most likely it's interconnections and, and all of that kind of thing.

So there's going to be, we should have expected that certain kinds of grammar or certain kinds of words or, or specific words would show up because that's what we've created, right?

And, and they'd, the foundational AI makers do have admitted that, you know, they put a, they put the thumb on their scale, you know, with their, their post training or their system messages and things like that.

But I think the, the most important part for this whole process or of discussion is the aspects of censorship.

We've we've censored ourselves in the way that we write.

Like Andy, you said, I, you know, I like using it or Beth has said, I want to continue to use the the M dash.

But because because there's a negative connotation to AI written material, it's used as a bar to filter out and to negate or put a stamp on it saying, Oh, this is not real, right?

It's not about the content, it's about how it's presented.

And I think that's a definite slippery slope because we should be able to express ourselves however we wish really, you know, And so if you like the M dash, but now there's going to be AI filters.

Oh, we don't want any AI generated content.

So we're going to use this as a filter.

Grammarly, you know, and, and other tools are going to have oh, let's make it less AI.

Let's make it sounding more human, things like that, which I always found funny is because the AI is just replicating what how humans write things.

So what are you doing?

As opposed to what I think's a better way to look at is this look at the content.

I think the people early on when people saw AI written material, it was just bad.

It was just bad writing that so that's what you should hound on is the bad writing, not the fact that AI made the, you know, wrote the actual characters that that's not really relevant from my perspective.

It's it's just like improve the writing and like we always say, you know, this is the worst that I'll ever be, you know, that kind of thing, right.

So, and then the other aspect of the, the censorship side of things is and the materials for, for the related, the related materials for this story, there's the story of algo speak right?

And where you're, we have changed our language and the way that we say things and the words that we use to represent, you know, certain meanings.

So that it's more in alignment with the algorithm.

Because the platforms in which we get to express ourselves on are, are throttled, throttled down in our reach for our words because we use, if we use certain words, that's not conducive to the platform.

And this is most likely connected to advertisers and things like that 'cause they don't want to be connected with, with tougher subjects, you know, And there are certain words that, that set that off.

And so like you use the term on alive, right, for, for death or, or anything that's related to that, to that.

And so, so I think this, this is sort of, this is sort of the slippery slope that we have to pay attention to because if we, if we censor ourselves to such a point where we're changing our language because it's to the benefit of a platform and not benefit to us, then we're just, it's just going to be more and more layers.

I just see this going down to this point where we're starting to use doublespeak and we're starting to obfuscate the the actual meaning of things.

Now admit, admittedly language does evolve, but that's usually out of a, a generational or natural evolution of that language.

Not, and it's usually to the the benefit of communication between, you know, either certain age groups or certain niche groups or subcultures and, and things like that, not for the benefit of a particular organization or a group of organizations, but that also explain also talks about how much reach that those those organizations have.

And so this might require more, you know, 30,000 afoot view of what's going on, but I think I I think the censorship side of things and how how we approach it.

So maybe this conversation about the M-IS just a symptom of people just seeing this immediate thing and not realizing that what they're doing here now is going to have a larger effect.

So maybe not quite take a take it, so make it make it such a big deal.

Well, I wanted to bring on Carl's with us.

And so welcome Carl to the conversation.

And I think this is a great time actually to bring you in, Carl, because, well, you happen to live in Canada and the other three of us live in the US.

And so while we all can understand each other, and naturally Canadians and Americans don't have too much trouble having conversations, there are things that are uniquely.

Canadian just like they would be for different areas of the United States.

In fact, I was just laughing about this and trying to find the origin of this.

When I was a kid for the one year I lived in California, we would play, I guess it would be called handball.

I I don't really know if that it's not handball because it was a giant like rubber bouncing ball with a giant like 20 foot wall.

I'm not sure if that's technically handball, but it's what we used to play out in the PE yard.

It's the only place I ever played it anyway, when you would get a point on your friends, you would very like aggressively go, oh, facial scrub.

I cannot find the origin of facial scrub.

It literally might have just been something one kid you created at that school.

And it does not seem to exist in surf culture or anywhere else.

Despite my best attempts to perplexity.

My answer perplexity is like we've looked, man.

It's not.

It's not there, you know, And so I only bring that up as an example because there's things that I've certainly heard my Canadian friends say that are more like familiar to like Canadians, right, versus something we might say in the United States is slightly different.

And so I want to bring up this idea of like basically digital colonialism, the idea that you kind of brought it up, Jimmy, when the in the fact of like UN alive or when you're watching tik toks and stuff, how people are using algo speak is what they call it to get around the algorithm.

So it's not like we're not already censoring ourselves in order to meet our goal, which is get around this algorithm.

Or I might put in prompting in order to get it to try.

It won't.

It won't.

It won't agree with me, but the try not to give it delves and M dashes.

And, you know, negation is something that I actually struggle with quite a bit because it'll do a lot of negation in its grammar.

So Carl, I just want to bring you in as somebody who's, you know, not in the United States, obviously in Canada.

So even in your unique experiences of using any of the models, not just shot GBT, have you found that it feels like it's sort of throwing everything into a soup where you're not maybe getting the type of grammar, the way that you would speak?

And I don't mean just like research or what you might show to a client.

I mean you just writing for the sake of writing or or sending something to your friends or writing a song.

Have you seen where it's maybe not lining up for you personally?

I think for Canadians, the the things, the things you probably see are anything that would be a ed would be replaced with an SI.

Can't think of a word right now, but we have words that are eds that aren't, that would be an S for in like APA style, I think or, or or just naturally for Canadian versus American.

And then the word OU versus just an O.

So shoot, I can't even yeah, yeah, yeah.

So something like that, right.

So when you're writing not, mind you, all large language models, they whether it's OU or just O it, it'll easily discern it.

But go, OK, that's what you mean.

But when the output is not that I care, but the output's always going to be O.

So it's not going to be Canadian, it's going to be an American, right.

So, but you can easily, you know, if you do your customization, you can do that and switch those things if it really bothers you.

If it doesn't, then you know, I've, I've had bosses in the past that would always like, you know, editing a whatever piece that you wrote for somebody or whatever the piece would be.

It would be like, you know, edit OU instead of O yeah, Especially like like very one of my very first jobs working for the federal government.

It's very important that you put the OU.

But obviously, you know, communication to Canadians has to be Canadian, right?

So those are the things.

Other than that though, I kind of agree with Jumi where eventually especially, you know, you can put in your customizations.

But even if you don't, as the memory and context gets better and better and like what we're seeing with like for me O3 where it remembers the conversation, I can, it's inevitable that it'll remember your writing style, how you write, how you think and all that.

So then the language itself will be reflective of how you, if you put these instructions how you want to output or the style you want to output, it'll output that.

I just, I think, I think the M-IS, you know, just part of the growing up of how AI is going to be part of our, of our like daily use.

And you know, I don't know if it was taken out, like blown out way out of proportion where I think like I think sometimes people make a big deal out of it because it's, it's a, it's a way to create your content.

So it's sort of like I'm going to make a big deal out of, let's say an M dash, and then I'm going to create content on it.

But then I'm going to make a big, then there's going to be the counter.

So I'm gonna create content on that.

So it's just to me, a way for people to create content.

It's like I'm gonna, what content am I gonna go put in my editorial calendar the whole topics of what M dash?

So I'll counter it with this and then I'll counter counter with this.

So to me it's like, OK, that's fine.

Well, I think we're just going to get past it.

And I.

I want to touch on the point that you raised, which is that, you know, you can provide custom instructions and you can provide a voice sample voice in the in the written form like this is the this is a sample of the voice of my personal writing.

You can give it that example and it'll follow that style.

And hey, eyes are completely fluent in every style.

I want to share an anecdote.

It's a Canadian anecdote.

I was in 2023 in February.

OK, so this is very, very early on.

I was up with my friends skiing at Whistler and I was trying to tell them about ChatGPT.

And so I pulled it out and I, and one of the things that I showed them was I said, OK, so my, my buddy up there is from Cape Breton.

Now, Cape Breton has a really strong dialectical form.

Like if you speak in Cape Breton, it's very distinctive.

And so I said to Chad CPT OK, I want you to, you know, write a little story about this this, but do it entirely in in Cape Bretonese.

And it did it.

And I read it out loud to them and they're like, I can't believe it.

That machine knows how to speak in that sort of almost comedic language at this point.

Like if you, you know, like a stand up comedian in Canada might speak that way in order to get a laugh.

Well, AI has every known dialect.

And so I want to give you another example.

I, I, I, I asked just just now, I said, OK, look, I want you to give me an example in standard American English of instructions on how to, you know, get to, you know, a highway.

And so it said, OK, here's standard American English quote.

I can see you're having trouble finding your way here.

The main highway you're looking for is about 3 miles north.

You'll want to head straight down this road until you reach the traffic light at the intersection, then make a right turn.

After that, drive approximately 2 miles and you'll see a large blue sign indicating the highway entrance.

If you miss it somehow, just ask anyone at the gas station on the corner.

They'll be happy to help you out.

OK now I said now do it in working class British English right then I can see your proper lost aren't you?

That motorway you're after is about.

Sorry, but I don't have the full accent right, but the word, the phrasing and the wording is is is so distinctly different.

The motorway you're after is about 3 miles up north from here.

What you want to do is carry on straight down this road until you get to them traffic lights at the junction, then turn right.

After that, go for about two miles and you'll clock a massive blue sign showing you where to get on the motorway.

If you somehow managed to miss it, just have a word with anyone at the petrol station at the corner.

They'll sort you right out, no bother.

There you go.

So you can ask AI to speak in almost any language.

Now it's going to default to a standard American English for explanatory purposes, and we're worried about that having an undue impact on the way people speak.

Well.

I, I, I'm skeptical about that because I think that people are going to speak idiomatically based on their cultural surroundings rather than what they're reading on AI.

So I I don't I don't think that that's going to bring to 1 dead level every mind.

Well, I, OK, I'll push back on that.

Unless, unless what Carl said comes true, which I, I think it would.

So my pushback would that be is people are inherently lazy.

And so I would think it's hard for me because I'm not, I'm not the ideal test case for this.

Chatsby does sort of already sound like me, you know, So I'm not, I'm not a great use case for this.

But if I was somewhere else and I had a different dialect over time, over hundreds, thousands of repetitions of using AI, we're whatever percentage we are today, Jimmy said.

Where it's the worst it'll ever be, it's the least it'll ever be, right.

So it's going to, that percentage is going to go up.

I would think if I don't know about Canada or I don't know Carl, you feel this way or whatever.

But if your kids now use your younger kids now use more and more and more AI and every single time it's giving them back a homogenized version of English that may not be what they would typically have learned in school or in their surroundings where you live in Calgary.

I would think over time that will just seep in it would it would almost inevitably have to that it would start to have an impact.

You know, maybe Carl's resistant to it, but his kids might not be.

And so like, but but what I would say to that is I don't know, Carl, maybe you're right that this is really a short term problem and that as memory and more personalization become more ubiquitous and become better with AI, that it actually undos, undoes, undoes this problem.

I say undoes, OK, it's the way I write.

Don't come at me.

I say I'm.

There in Florida, that's.

Going to be my new every time I mispronounce a word on here, I'm going to be like guys, it's the way I say it.

Don't try to change me AI.

It's a good, it's an easy out undoes what?

What muscle trading does this thing?

But anyway, my point to it was maybe that undoes this current trend, this homogenization, this digital colonialism, this need for algo speak.

Well, algo speak may still be there because the algorithms are going to have words like the opposite of the other word for on alive that could trigger this exact YouTube video.

We're live on YouTube.

And so we could get demonetized.

We could get penalized on this show for saying certain words.

And so not that we think about it a lot, but I do think it's part of the conversation that we would have to change the way we speak naturally having conversations on the show.

Because somewhere in the back of our brain, if we were a much larger show on YouTube, we were a much larger channel, that's literally dollars in our pocket, literally because it could demonetize us.

And that could be a huge deal.

Like if we're not not getting the volume we have right now.

But if we were getting a million typical viewers on one of our shows, that's a big chunk of change.

And it's definitely going to change the way we speak because we're going to work our way around the algorithm in order to make sure we don't hit those roadblocks.

So is this just a temporary thing or is this actually something that's going to this AI speak, if you will, this this always to the mean?

And when you also bring in the idea of what the algorithms are trying to do and how they work efficiently, when you bring that all together, is this something where we look back in 10 years and go, we should have been more aware of this and now it's actually caused cultural problems where it's changed fundamentally the way people speak.

And we kind of wish it hadn't done that.

Or is this a short term problem because AI will actually solve for it with higher levels of personalization that will wipe out these effects?

I I think like the, the, I think to answer that question too, what I would, what I'm, what we, I think the there's other factors related to that would be it depends on how, how much does your, let's say your children engage with the AI versus engage with each other, right?

So like I can hear my daughter and her best friend, which is a a boy and they're like bra pretty much the entire time, right?

There's not, but I'm not, you're not going to have that with AI.

And like Bra, you're like the way that you converse with your group of friends versus you converse with your AI.

Like I don't, I don't know if there's that delineation between I'm going to speak exactly how I speak to my friends in different settings to how I would speak with an AI.

Mind you, now I use things like a whisper flow and just just talk to it, right?

But I still don't talk to it as if I'm talking to my fantasy Football League, right?

I kind of don't talk to it that way.

Maybe I, I, I start doing it, but I don't.

But I just feel like the way the dynamics you have with your people and your friends growing up maybe more dictates how you would engage with this rather than the other way around, that it impacts the way you engage because it unless over time, like how you would talk like you, you talk more and more to the AI versus real people.

And, and maybe for some people that is that's the case.

Yeah, I, I think you, you definitely highlighted a a part of what I think the larger or the my, my answer to your question Brian is it's an additional problem.

It's not something to be to be solved.

If if you look at from the framework of if the goal is to homogenize and have a standard way of speaking and thinking and understanding across the world, then it's going to be successful at that.

Because the more you interact with something, it's a new tool.

It the, the more you interact on that certain levels just like, but when people have business speak right or the way that they speak in a business, English is the standard commercial, the global commercial language, right?

And there there is tons of programs across the world to help people learn English to conduct business globally.

AI is, is a new tool just towards that kind of effect, right?

And Carl, you mentioned, you know, interacting with the AIS more.

I think we one of the things to pay attention to is, is how much interaction between people and how much interaction between AI as AI becomes more and more developed, like character, not AI and more interactive and you're able to have those more extensive discussions or conversations and things like that.

So that's, that's definitely another add on.

So I think, I think with most things we tend to as people, we, we'll see what the immediate thing is.

And if we see it as an immediate problem, we'll deal with it right then.

If we don't see it for something that's going to happen in a decade or a couple of decades or a generation or two down the line, we, we tend to not see that far down the tracks, right?

We tend to see what's right here, right now.

Oh, that's a problem I can fix right now.

If that's down the ways, I will pay less attention to it now, unfortunately, with very large shifts, they tend to take time.

And if you're not paying attention to them, they're going to shift, you know, in, in that direction.

So I do agree that we, we need to pay more attention to every, every step of the way.

I know it is exhausting and it's tedious, but you have to because the one thing we can, I think we can all agree on is like you said, AI is immensely powerful.

It touches every aspect of life and society and it changes extremely fast.

We don't have the benefit of, of, oh, I'll look at it in, in 10 years or I'll, I, I'll, I'll do some small steps, right?

It'll, you know, before there's literally, if you had a kid now and before they became an adult, AI will have completely changed 1000 times.

So, you know, so it unfortunately we we have to think harder and longer and pay more attention which we kind of have been trained not to do that.

And by the way, I want to bring up that there is on that point, Jimmy, there is research.

I didn't bring it up at the top of the show, but I'll bring it up now.

There's been research done on this.

This is not like, you know, my idea or our idea that maybe perhaps this is happening.

No, it is because there was a study done Max Planck Planck Institute.

I think that's how maybe you say it for human development provided combelling evidence of this trend by analyzing a massive data set of over 771,000 podcast episodes and 360,000 YouTube videos created between 2019 and mid 2024, researchers identified a statistically significant increase in the use of specific AI favored words following the research HPT.

The vocabulary in question included words like delve, meticulous, adept, realm, tap, industry, nuance, prowess, bolster, and streamline.

These words, previously uncommon in everyday English, have seen a dramatic rise in frequency.

For instance, in a study of academic YouTube channels, the use of the word delve jumped by as much as 51% after Shatsopati became publicly available.

So this is happening.

It's not like maybe, maybe it is.

It's happening in the background.

No, no, there's studies and they've been done.

And you know, it's not just these words, it's other ones.

And these are the ones we.

We see and have come or have risen to your point, Carl, about people creating content.

It becomes this, it becomes this echo Chamber of like Delve.

And then I make a, a, a, a post about why I hate Delve.

And then I do the counter argument.

You're right about that or tapestry.

I think that one is in there.

I see that one quite a bit.

Who's I?

I never said tapestry, you know, before that.

So then it comes up in the answers and then that becomes public talking points and social media.

These are just the ones we're more familiar with.

I, I would imagine there's many, many other smaller nuances of the way AI speaks.

Like I said, the negation that like where what I mean by that is like when it starts a sentence and it says this isn't only happening now comma it's the future.

That is such an AI unique way to speak that I will undo it anytime I see it and my work is going out publicly to a client.

So to me, it's that we're I'm beyond one words.

I mean, yeah, Delve M dashes, they annoy me.

Sure.

I'm now at parts and phrases of the grammar that I'm having it actively undo.

Because if you say this isn't just something that's today, this isn't just, this isn't just fantasy, this is reality.

You can't tell me that wasn't written by AI, even though three years ago I would have said that's just a normal sentence.

People do that all the time.

They use, you know, passive versus active verbs.

I, I learned very quickly in my master's program that, man, do I like some passive verbs.

I couldn't even told you what a passive verb was before I started my master's program back in like 2006.

But my wife, I had like, you know, whatever the word add on was at the time, I didn't know AI.

My wife would read all my professional papers that I had to submit and say, you're still using passive verbs even after my attempt to try to get rid of them because I had to do whatever the standard was.

The grammar standard was that it started with an A.

It's an acronym that everybody had to use for, for all the professional writing I had to do in my master's program.

And you could not have passive verbs.

They had to be active.

So all that had to be changed out.

So anyway, I just bring it up as like, yes, we've been doing this, but yes, there's, there's research on this that's happening all the time.

And so the question becomes, will AI enable us to bring back our personalization?

Or as Cisco pointed out, hey, it really depends on who my end target audience is.

If it's my wife, my kids or my job, my boss, or maybe government writing that dictates how I will speak to the AI.

And maybe that's enough.

Maybe that will be enough to, you know, quell this.

But I do worry that it's one of these silent type things where we're going to certain communities and they're going, no, it's, it's not just bra, it's not just 6-7, It's not just whatever the kids are saying today.

That is just the flavor of the month and will be gone next week.

You know, it's like, no, it's actually hurt our, our language or hurt our dialect or something like that.

And that that makes me worried.

I'm glad you I'm glad you mentioned the latest dialect of the moment, so I want to take a comedic interlude here.

I'm going to share my screen and this is just a short it's a YouTube short, but listen to this.

This is going to have a much larger impact on the next generations forms of language than AI will.

I hope AI will have a larger influence than this listen.

Can't hear it?

Yeah, that's what's happening.

Andy can't hear us.

We can't hear it.

OK, I'm going to hide it on stream because we can't.

We can't hear it at all.

There's, there's one thing I want to tell you to add to this too, because I think what I'm wondering is like in the workforce, will certain businesses have their own language that you'll have to adapt to cause or how they use their AI 'cause this is the thing that we're thinking of, right?

So it's like, hey, each of us will have our own agents, personal AI, personal super intelligence, whatever that is over time.

But then wouldn't the businesses that you work with have their own, Wouldn't they want, or will it become so that, hey, we want either we want your AI to match how we do things, the way we do things, language included, Or does the business kind of just take AI as, you know, a collection of its employees and add it to its, you know, how it operates?

So I 'cause I was thinking about that, it's like if you're bringing your own AI and all the history and experience that you've worked with it, and this is probably a little bit further on, would, wouldn't you take all the like, for example, we have now our agent mode.

We have our own claws or Geminis or Chat GPTS and it's super, I'm, I'm noticing it's tailored to us.

It has the memory of the things that we've done.

Now when we bring it into a work environment, it's still tailored to us.

Will the businesses eventually say, hey, will you either want you to use our version of these models because we've set the customized, we've customized it for our business or do you allow it to say, hey, yeah, bring your model in.

You know, we'll have some security measures.

But the whole point is we, we hired you for your experience and your AI for its experience and history.

So that's the question, because what I'm running into is some of the work we've done with clients because of chat GP TS history.

It's referring back to some of the work we've done with previous clients that I'm starting to think, wait a minute.

If I, let's say I use a brand new model or have a different ChatGPT account, let's just say that it will not have some of the context that my current account does for all these clients that we work with.

Because honestly, I'm starting to put, Hey, do you remember what we did here?

Can we also apply this but this and it remembers what we did with another client.

So but I would, but I can't do that with something else to the point where it's actually starting to get a little bit and a little bit where, oh, oh, I'm now relying on it.

It includes the memory capacity.

It, I, it, I'm leveraging everything that we've done with previous things because I can't remember all the things it's done, but it, it, it, it does.

And it's like, Hey, do you remember what we're doing here?

Can we apply this?

But it's like, Oh, that's a good idea.

It's like, ah, so that's what I mean.

It's like if we're doing a lot more of that then because imagine Brian, all the work you've done building all these things, all these things that you've worked with different clients with, imagine you having to, you switch jobs, but you're not allowed to bring whatever you did.

And it's the same, the exact same job or higher job, kind of starting from scratch, kind of kind of nod, right?

Like it's a little bit you're like, oh crap, I have all my custom GPDS.

I have all the instructions, everything, but a little bit of the context is missing.

And I'm curious to see like can, can, can you keep that context?

Like is there somewhere I can take it and bring it with me?

I, I don't know.

I I I don't think so.

Well, I mean, well, OK, so I'll answer that really quick before we start wrapping things up.

Scaled Jake, my CEO, it's not really through scaled, it's not launched yet, but it's called Journey AI.

And funny enough, that's one of the problems it's it's trying to solve for Carl, which is that sales reps often work across multiple different clients.

Like they're, they're, they're talking to multiple, I'm not clients, I'm sorry.

They're talking to multiple different like lead Gen.

and prospects and stuff, but also they sometimes move and because like it's not uncommon for a sales Rep to maybe change companies that they were for two or three, four times before they move up the chain.

It's pretty common.

You're not, you're not staying with a company 20 years usually in that in that position because it's actually advantageous for your career if you bounce, right.

So in that case, you're right, that Rep loses a lot.

Now, could they reference a previous client across to a new company?

No, we know that.

But there's a lot of that memory and context that's left lost.

This is akin to back in the day before a cell phone would transfer to the next new cell phone.

It was painful to leave an old cell phone back when they look more like beepers back in college days when it was just a few little digits and you had to press, you know, 3 three times for the letter, that kind of cell phone.

But you guys remember, there was no transfer knowledge.

There was no universal oh, your new phone magically knows all the stuff that was on your own phone.

Same thing with computers.

You just lost it and you sort of started over and AI, you're right, is going to solve for that over time where you no longer, you know, when I was a kid, I would carry your own 1010 to 15 phone numbers of my friends on a little piece of paper in my wallet.

Well, then the, the, the cell phone solved for that.

Unless you lost your cell phone, then you were in a world of hurt.

And now I don't really have to worry about that because those phones are going to be carried on to the cloud and it's not really an issue in mind.

And so we, we've gone through these pains before.

And I think AI is going to solve for that.

But you're probably right to the point of you will get so comfortable and it will be so good at its recalling of memory that you will no longer carry around the the theoretical or not theoretical, but the hypothetical pieces of phone number on a piece of paper anymore in your wallet.

You won't have to.

You will.

You will believe that the system will take care of you, and the technology is so foolproof that that recall of that AI memory is there for you when you need it.

One thing though, and I'm thinking about this.

So a lot of the work that I'm doing, whether it's in Claude, whether it's in chat, GPG, whatever, but we all know certain models are better than others at certain things, right?

Despite whatever all the the providers are telling you, some models just better than others at certain things that they want to do.

Technically, I want to be able to take the quote UN quote context and the memory of what I'm doing across the board.

Not I have to re explain this context to like clawed.

I have to re explain this context to a copilot.

So let's say I'm working in one company and I have this context.

So I don't know, Let's say I'm in HR, have this all this HR things that I'm doing.

I'm not going to take the like the private information and the specifics to that business, but what I want to do is take all the experience, all the ways that we've done it and take that with me to another company.

So I'd be like, hey, do you remember when we worked for Company A?

We did this.

I'd like to do like I may remember it, but I want all the AI, the threads and all the conversations that everything that we've taken from 1:00 and bring it with me to another.

And I think that's a it's a very interesting and I don't know if this if this has even crossed anybody's.

I'm sure somebody's thought of it where a lot of people are working very, very heavily into one with one model, but from AI mean just from pure business.

And some there's, I'm sure there's a lot of other uses for this, but I want to take that and move it to another business.

Whether you're an employee, whether you're a founder, whether you're just personal, like how about this?

I want to also pass everything I learned or my AIS learned to my children, or I want to pass it on to family members or I want to pass it on to other employees.

Like how do we start moving that?

Because right now you kind of can't do that like.

So yeah, yeah, why?

Why isn't this?

Why isn't this practical using chat history in a rag system?

If you I think it's not practical because of the security and IP concerns from the company.

If you're an employee, this becomes very murky in terms of we want you to use our systems.

Anything that you derive or make while you're here belongs to us.

So we're going to have to filter and go through everything.

The reason why you can take experience is because it's stuck in your brain.

It can't extract it from your.

Brain, but that's But that's what I.

Mean, right, yes, I, I think your propose what you're talking about is, is going to be very important to consider how it's done.

And if the concept is we're bringing our AI companion or our IAI twin or second brain is a way for us to record our experiences, figuring out how a company can interact with that, with that source and still remain secure.

Is it is going to be something that that needs to be solved.

Because the big difference is I think the first step will be if you're an employee or if you're, you're a contractor.

If you're a contractor, then you can say, hey, this is my system.

You can go whatever.

But if you're an employee, you have very little wiggle room.

Well, that's what I mean.

It's like I'm not hiring just you Junie or Brian or Andy.

I'm hiring your AIS with you that I'm assuming.

Yeah, right.

I'm assuming down the road everybody's AI will have everything that because I want to be able to leverage all the experience that you and your AI has.

100% it's like your it's like your personal brand.

You know, Carl, if you have if you talk about a specific industry and sub industry as a contract as as somebody at like zero to 60 and you have a million followers on YouTube, that is valuable to the next company for sure because they're thinking, well, Carl will now have that logo at the bottom of whatever.

We're not just hiring Carl, we're Carl hiring Carl's personal brand.

So yes, and the in the future, I'm not just hiring Carl, I'm hiring Carl and all his AI assistants that he's built and cultivated over time.

I'm not hiring one person.

I'm I'm hiring 100.

Right, which and that goes towards.

Everyone becoming their own their own entity.

But I, I think what we just unlocked here, and I think probably a new show here or something, is that this discussion has to be, well, how does that work now for a company, right?

Because all the processes and systems in HR and all that stuff is all centered around just a person.

You're hiring that person, you're hiring that person.

It's like, well, no, you're not.

You're hiring that AI, you're hiring that person's AI.

And I guarantee that person's AI will want to connect into your system, not the other.

I I want you to use ours like no, no no.

And then the question about who retains.

What like that's mine it it requires radical reassessment completely on your systems that and companies will have to have to decide to go down that path or or it's just not gonna happen, but.

OK, I'm gonna I'm gonna wrap this up only because I know I have to run and I just worry about I'll kill the feed on this as I'm the one that started it.

Background problems for the show Real quick tomorrow we have the news with Jimmy Wednesday we're going to be talking about AI and aging in place with dignity.

That was brought to us by Diane, who we really appreciate.

We'll talk more about that, but that was a recommended show.

Friday, Friday, Friday, Friday, it's our two year anniversary show.

88 is when we kick things off.

And so on Friday we're going to have our two year anniversary show and we're going to have all sorts of fun conversations.

That's what's coming up the rest of the week.

Make sure you like and subscribe if you like what we're doing, we're here here.

We're here every weekday Monday through Friday from at 10 AM Eastern live.

So you can always check out any of our 521 other shows.

This is 5/22.

So that's it.

We'll wrap it up.

Great conversation.

There's more to come on this for sure.

Thank you Greg.

You said great conversations is what we appreciate all of y'all in the in the comments get better feel well Cisco who's fighting a bit of the flu and we will see you guys tomorrow for all the amazing news of the last seven days Till then bye everybody.

Aloha.