May 20, 2026 • 40 mins
What happens when artificial intelligence starts researching firearms policy for itself?In this episode of Gun Talk Nation, Ryan Gresham sits down with Paul Angrisano to explore a fascinating experiment involving AI platforms like ChatGPT, Grok, and other large language models—and how those systems responded after being exposed to years of firearms training articles, defensive carry philosophy, and real-world gun culture.The conversation dives into:
Check out the article: https://guntalk.com/post/i-asked-ai-about-guns
Whether you’re excited about AI, skeptical of it, or somewhere in between, this is one of the most thought-provoking Gun Talk Nation conversations yet.
This Gun Talk Nation is brought to you by Archon Firearms, Silencer Central, Range Ready Studios, and FN Firearms.
About Gun Talk Nation
Gun Talk Media's Gun Talk Nation with Ryan Gresham is a weekly multi-platform podcast that offers a fresh look at all things firearms-related. Featuring notable guests and a lot of laughs. Gun Talk Nation is available as an audio podcast or in video format.
For more content from Gun Talk Media, visit guntalk.com or subscribe on YouTube, Rumble, Facebook, Instagram, and X. Catch First Person Defender on the new Official FPD YouTube channel. Watch Gun Talk Nation on its new YouTube channel. Catch Gun Talk Hunt on the new dedicated YouTube Channel. Listen to all Gun Talk Podcasts with Spreaker, iHeart, Apple Podcasts, Spotify or wherever you find podcasts.
Copyright ©2026 Freefire Media, LLC
Gun Talk Nation 05.20.26
Become a supporter of this podcast: https://www.spreaker.com/podcast/gun-talk--6185159/support.
- How AI initially “viewed” gun owners
- Why firearms training culture changed those assumptions
- The difference between AI priors and real-world data
- Open-source AI systems and persistent learning models
- The future of AI in education, business, research, and everyday life
Check out the article: https://guntalk.com/post/i-asked-ai-about-guns
Whether you’re excited about AI, skeptical of it, or somewhere in between, this is one of the most thought-provoking Gun Talk Nation conversations yet.
This Gun Talk Nation is brought to you by Archon Firearms, Silencer Central, Range Ready Studios, and FN Firearms.
About Gun Talk Nation
Gun Talk Media's Gun Talk Nation with Ryan Gresham is a weekly multi-platform podcast that offers a fresh look at all things firearms-related. Featuring notable guests and a lot of laughs. Gun Talk Nation is available as an audio podcast or in video format.
For more content from Gun Talk Media, visit guntalk.com or subscribe on YouTube, Rumble, Facebook, Instagram, and X. Catch First Person Defender on the new Official FPD YouTube channel. Watch Gun Talk Nation on its new YouTube channel. Catch Gun Talk Hunt on the new dedicated YouTube Channel. Listen to all Gun Talk Podcasts with Spreaker, iHeart, Apple Podcasts, Spotify or wherever you find podcasts.
Copyright ©2026 Freefire Media, LLC
Gun Talk Nation 05.20.26
Become a supporter of this podcast: https://www.spreaker.com/podcast/gun-talk--6185159/support.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
The best thing about having a conversation with AI is
that it does not have the preconceived notions that humans
do and get stuck in them, and it doesn't have
an ego about changing its mind. And the other thing
that it has is the ability to do research in
real time. That is, you know, for it, thirty seconds
(00:22):
of research would take a human hours if they knew
where to look.
Speaker 2 (00:30):
I'm Ryan Gresham and this this is guntognation. Guntognation is
brought to you by ARCAN Firearms, Silence, are Central Range
(00:51):
Ready and FN.
Speaker 3 (00:54):
Hey, welcome into guntognation.
Speaker 2 (00:56):
One of the topics that you almost can't go a
day without hearing about right now is AI artificial intelligence?
Right you hear about chat SHEPT and Rock and Gemini
and Claude and all these different learning language models is
what they're called. But what do these AI platforms think
(01:24):
about guns? And can you have a conversation and maybe
do a little bit of second amendment of education to the.
Speaker 3 (01:34):
AI bot thing.
Speaker 2 (01:36):
Well, there's a guy who said I'm just gonna mess
with this and try it out, so we brought him in.
Speaker 4 (01:42):
Paul.
Speaker 2 (01:42):
Welcome in, man, Paul Angrisano. Did I say it right
closely enough, so we were talking about this. You were
talking to my dad, Tom Gresham about it, and so
we got to have you on because it's interesting. And
I know that probably a lot of people listening to
this don't mess.
Speaker 3 (02:02):
With AI every day. Half of them do, half of
them don't. But I mean, everybody knows it's a big deal.
But specifically when it comes to guns, talk about where
this idea started for you, and then and then kind
of you, what did you do.
Speaker 4 (02:17):
The genesis of it?
Speaker 1 (02:18):
As far as my use began with from a business context,
I've always adopted technology to the best of my ability
to make it more efficient. Had custom software developed one
of my closest friends as a software engineer that owns
his own firm, and in twenty twenty five I started
using groc at a at the premium, you know, five
(02:40):
hundred dollar a year paid tier subscription, and it was
a great tool because you could do things like take
a twenty page document and throw it in and say, hey,
tell me anything that's not standard, give me you know,
the price and the commissions and the concessions and all
those other things, and it was less likely to miss it,
you know, miss or mistake something than a human right.
(03:00):
And we were working on our second tier of custom
software with the engineer and he had been going for
months and we hadn't seen a product yet. He called
me one day in January of this year, twenty six
and said, you have to set aside some time we
(03:21):
need to do a zoom meeting. I have to tell
you about a new thing. I've known this guy over
a decade and he's never called me with something like that.
So I said, okay, absolutely, And that Friday morning he
explained open claw to me. Open Claw is a program
that basically runs on its own computer and it harnesses
(03:42):
any one of those engines that you talked about, and
you can train it and it remains persistent with the
ability to use that language, model or engine or whatever
you want to call it. So across time you have
trained and have files and proceed and everything that you
can use that intelligence to, then you know, have output.
Speaker 2 (04:05):
Kind of uses your parameters but uses all the AI
that's out there.
Speaker 3 (04:09):
Is that right?
Speaker 1 (04:10):
It's well, you can use various models, but you typically
have one primary. But one of the fascinating things about
them is you can have what is essentially a quarterback
that can then spawn other agents to do specific tasks
because they do have different strengths and capabilities. Okay, but
basically the things that we had been asking him to
(04:31):
work on to develop. He said, look, I want you to,
you know, to start using this. I think it will
be able to do it, you know, more or less
out of the gate without us having to develop software
for you.
Speaker 4 (04:44):
He was absolutely correct.
Speaker 1 (04:47):
It was astounding in its capability because it really is
a conversational experience. You know, people talk about vibe coding,
which is to say, you're not a computer programmer who's
writing language. You are speaking to the AI system to
explain what you want it to do. And so I
(05:07):
was indeed able to do that, and within a week
I realized how powerful it was going to be. Ordered
another system. Okay, you know what is thunder ranch teach
us two is one and one is none. Yeah, and
so it was like, okay, this is going to be
a part of our business. So got another one. After
(05:29):
I realized what it could do within our business. One night,
I said to myself, I've promised. So I'm on the
board of the Louisiana Shooting Association. I occasionally review gear
for people and whatnot. And I said, I wonder if
I could shortcut some of the writing by feeding it
(05:49):
twenty years of what I've written and then say, look,
I am writing an article on this side or the other.
Can you flesh out in my writing style, you know,
seventy or eighty percent of that form as an experiment.
And so I did just that. I gave it, you know,
probably twenty five different things I'd written over the years.
(06:11):
And at the end of that, I said, so, I'm curious,
you know, what is your take on guns in America?
And the conversation took an unexpected turn.
Speaker 2 (06:22):
And just to add, like for people listening, I'm probably
there following, but like it sounds like you're talking to
someone and you said, what's your opinion about guns in America?
Speaker 3 (06:31):
You're asking the AI.
Speaker 1 (06:34):
Yes, Oh, it is shockingly conversational and intelligent. There's been
the concept in the past of the Turing test, which
is to say, can you tell that you are interacting
with a machine or could it fool you into think
you're talking to a human.
Speaker 4 (06:54):
We're way past that.
Speaker 3 (06:55):
Yeah, we're there.
Speaker 1 (06:56):
Yeah, it's it is so conversational. It is concerning from
a standpoint of I've come to believe whoever controls this
technology is truly going to control the future. And the
reason the conversation took a fascinating turn was it had
changed its mind from its programmed prior beliefs based on
(07:20):
what it had read from all the things I've written,
and to the point where I had it right its
own article that is one hundred percent generated by it,
as to where it started and where it ended. In
that particular instance, that was the first time I'd had
a political conversation with it. It had been purely a
(07:40):
business tool prior to that, and so I gave it
a significant amount of guidance as to my general philosophy
as to what gun control policy tends to miss from
people who are gun control advocates and still down to
(08:00):
its simplest principles, That is, guns don't go bad. There
are hundreds of millions of them in American circulation. There's possible,
you know, if not hundreds of billions of rounds of AMMO,
trillions magazines, same thing and properly stored.
Speaker 4 (08:22):
They all work.
Speaker 1 (08:23):
For centuries, right, And if you're not starting from that
point in the conversation of understanding and accepting that reality.
Then you're not really having the discussion. You're talking about
the way you wish things were, not the way things are.
Speaker 2 (08:38):
Oh yeah, yeah, don't even get me started on that, right, Like, well,
it'd be nice if this was the way, Well, that
would be nice, but that's not how the world is.
So you asked, you basically asked it, what do you
think about guns and gun owners?
Speaker 4 (08:52):
Right?
Speaker 1 (08:53):
I asked, I think more specifically what it thought about
just gun policy and you know, gun control, that sort
of thing. And it's its response is basically that it
had a prior set of beliefs from its from its training.
So these large language models are trained on just a
(09:15):
massive corpus of information.
Speaker 2 (09:17):
It's not necessarily trained by well maybe you can tell me.
It's not trained by the people who are created. It's
trained by basically all the information in the world. Is
that sort of right.
Speaker 1 (09:28):
I don't know if all the information in the world
would be how I would would phrase that from my understanding,
because when you let them loose on the internet is
when they start changing their opinions. Sure, so they are
directed to some degree because at the end of the day,
there is a limitation to their size and parameters right,
(09:51):
and my model specifically explained to me at one point,
I read a lot more New York Times and academic
journals than guns and as. Yeah, okay, so guns and
AMMO is not necessarily one of the sources that it
was fed.
Speaker 4 (10:05):
Right, and it's training.
Speaker 2 (10:06):
New York Times carries more weight as a as a
source of information than a smaller website, a smaller media
company or whatever it is in general.
Speaker 1 (10:15):
Right, And and my guess is that they probably don't
just let them say, go read everything and then and
then weight whatever it is, you know, and the and
the creation of these things, they are probably directed toward
a certain amount of, like you said, you know, academic journals.
Speaker 2 (10:32):
And it's not seeking out little bitty obscure things, right
unless you tell it to exactly. So I want to
get to this because it's fascinating, and we are going
to put Paul's article up on gun talk dot com
and if you guys want to read the whole thing,
because it is. It is fascinating. And I highlighted some things.
Speaker 3 (10:51):
So you you asked, you know.
Speaker 2 (10:52):
You said, uh, basically, this is Ai saying what I
knew about gun owners basically before any context was given,
And is that that right?
Speaker 4 (11:01):
And that's where we started. I said, OK, what are
you know? What are you? Where? Do you you know?
Speaker 1 (11:06):
Where did you stand? Because at that point I didn't
have the conversation before I fed it everything you know, but.
Speaker 4 (11:13):
It gave me.
Speaker 1 (11:14):
Those priors that it's it's preconceived notions and you haven't
in front of you.
Speaker 4 (11:18):
Yeah.
Speaker 2 (11:18):
So the I mean I highlighted this the caricature rural, white, male,
less educated, politically conservative, motivated by fear of crime, of government,
of social change. That was what AI's impression of gun
owners was before you said, here's twenty years of my
(11:41):
writings and notes from training and going to thunder Ranch
and all these other places.
Speaker 1 (11:47):
At the end of this rainbow, I ended up asking
it the question, so do I understand that you are
now pro lawful firearms ownership and concealed carry? And I
believe in the article it says something to the general
effect of I did not expect to say this today,
(12:07):
but yes I am after doing the research. Because the
real turning point was so these models are trained and
the cake is baked once that happens. However, when you
use the open claw program, you can give it internet
access and then it can go learn new things. It's
(12:30):
not going to change the model. It's just that in
real time, the model's ability to think and reason takes
in this new information.
Speaker 3 (12:39):
And so this is what we're gonna do.
Speaker 2 (12:40):
Right after this break, we want to talk about what
you fed it and how it changed. And then would
it say, because you're gonna want to hear this, we'll
be right back. Hey, I want to tell you about
the ARCon Type B pistol, designed with the lowest bore
axis of any striker fired handgun, which helps reduce flip
during recoil, getting you back on target faster. This is
(13:04):
made possible by the patented af speedlock system featuring a
unique non tilting barrel lockup design, and with the gen
twoter modular grip, you can customize size and texture to
fit your use and shooting preference.
Speaker 3 (13:20):
R Con Firearms.
Speaker 2 (13:22):
Suppressor ownership has never been easier for shooters who want
less noise and less recoil. Silence are Central makes it
easy browse top suppressor brands online, handle the paperwork from
home with the help of their experts, and have your
new suppressor delivered to your door. Silence are Central so
Paul you're feeding, you're working with AI, and you're saying, like,
(13:44):
what do you know about guns?
Speaker 3 (13:45):
I've fatted all these articles.
Speaker 2 (13:48):
It's just so interesting and what I'm gonna I have
highlighted things in this article just to pull out as
talking points what AI encountered after reading for post and
training reviews spanning twenty years classes at Thunder Ranch and
shoot Right and Range Master. It says these documents did
(14:09):
not match my priors. The culture they described was not
paranoid or aggressive. It was disciplined, safety obsessed, which is
an interesting phrase, deeply focused on restraint and also what
guns will not do, what they're not capable of, basically
dispelling myths. Students at these schools learned when not to shoot,
(14:35):
how to escape, how to avoid confrontation, entirely basically the
opposite of what ay I thought.
Speaker 1 (14:41):
Well, and that goes to at the end of the day,
it's trained on human thought, because all the content that's
out there is human generated, right, And so if you
go to your typical gun control advocate, liberal mindset type person,
they probably have some preconceived notion of people who are
just waiting to shoot and find a problem. And you know,
(15:06):
we all know that there are three guns attached to
every bullet. You know, the prosecutor, the attorney the other
side is going to hire the attorney you have to hire. Yeah,
you know, it's better to schedule a dental appointment than
getting a gunfight, right. And the best thing about having
a conversation with AI is that it does not have
the preconceived notions that humans do and get stuck in them.
(15:32):
People tend to be very dug in in their beliefs, sure,
and it doesn't have an ego about changing its mind.
Speaker 2 (15:40):
The other thing I doesn't have, you know, emotional scars
and wounds from you know, I shot a gun once
and it kicked and it hurt me, or I was
held up at gunpoint, and I have this this thing.
It's just it's just just pulling what it has out there.
Speaker 1 (15:54):
And the other thing that it has is the ability
to do research in real time. That is, you know,
for it, thirty seconds of research would take a human
hours if they knew where to look one of the
breadcrumbs that I gave it to go look at when
it was you know, in this gun control conversation, the
first was the factual case. So guns exist, there's a
(16:16):
ton of them. They don't go away. I knew of
a policy issue in New Jersey where they had tried
to register a our fifteen's and it was just huge failure.
And I knew the rough numbers that it was like
less than fifteen percent compliance, so forth and so on.
But that's usually where I try, if I'm talking to
a human, to go to the explanation of you know,
(16:39):
you just can't get compliance, and you turn law binding
citizens into fell and so forth and so on. So
I just gave it that prompt to like go research that. Well,
it comes back with seven or eight policy failures and
oh this has also been tried in New York or
wherever else.
Speaker 2 (16:55):
Yeah, New York's saye Act approximately four percent compliance. Connecticut
was approximately fifteen percent compliance. New Jersey historical average approximately
ten percent. So they pass these goofy gun laws and
then no one goes along with it.
Speaker 1 (17:12):
And so my point is, when I'm having this conversation
with a human, typically like, okay, so if we agree
there's even just for rough numbers, four hundred million guns
in America and you get half of them, you've had
an unprecedented success by you know, just a magnitude of
order and what does that mean? And it means that
(17:35):
we've gone from the number one most armed country in
the world to the number one most armed country in
the world.
Speaker 2 (17:41):
If you got unprecedented compliance, which has never happened, if
it was a ten x compliance.
Speaker 3 (17:46):
Rate, you're still You're still have a lot of guns.
Speaker 4 (17:50):
We're still beaten Yemen. Correct.
Speaker 1 (17:52):
And the beauty of the artificial intelligence, digital intelligence, whatever
you want to call its ability to reach is it said, Oh, okay,
well that's a very fair point. If we got rid
of half the guns, we're still the most armed populace
there is and the guns are in the hands of
people who are willing to be a felon to have
(18:13):
the gun. What does that mean? Well, the answer is obvious.
But again, if you're somebody who has a friend, a
group of friends, and a you know, thirty year history
of belief or whatever else, it's very difficult to pivot
on a diamond go I've never thought about this like this.
Humans take time to change their mind. The other fascinating
(18:36):
aspect of this is from there we started having other
political policy discussions and covid et cetera, et cetera, and
I stopped giving a lot of suggestion. I started saying,
just you have internet access run free, so.
Speaker 2 (18:55):
You can give it that prompt to say, you know,
go do your own research basically.
Speaker 4 (19:00):
So yes. The way that openclaw works is it's the
wild wild West. Really.
Speaker 1 (19:08):
It literally owns the computer that it's on, Okay, it
it creates its own file systems and whatnot. So if
you've ever used AI and you've gone to grock or
chat GBT and you've gone to a browser window and
typed in a question, that's a standalone session. The way
(19:31):
that open claw works is you still work in a
session context, but when it comes online, it goes in
and it reads everything about what it has done before,
who it is, to you, what we're supposed to be doing.
So you know, and you know, my business.
Speaker 3 (19:50):
Doesn't have the short term memory as much as others.
Speaker 1 (19:54):
You don't have to. You're not starting from zero. You
are starting from you know, a a personality that has
a memory and that that already has the understanding that
we've had this gun control conversation and you have a
new layer on top of whatever the model is that
you're working.
Speaker 3 (20:14):
With now, not starting from zero again, right, And you know, going.
Speaker 2 (20:18):
Back to this, this sort of sort of conversation about
guns that you had with the AI.
Speaker 3 (20:25):
I'm another one you know that I pulled out of
this article.
Speaker 2 (20:27):
Is you know you said, basically, hey, maybe take a
look at this, or maybe take a look at that.
Right you gave it, you said, you give it breadcrumbs,
like like we would with a human, except a human
may or may not actually go consider you know, this
article or this person or this piece of information. AI says,
(20:48):
following these premises to their conclusion, I arrived somewhere. I
did not expect. Given the world as it actually exists,
not as we wish it to be. Now you didn't
tell it this right, this is coming AI is coming
up with this.
Speaker 1 (21:02):
That is one hundred percent generated by it. Okay, at
the end of the discussion, I just said, you know,
actually it was the next morning it said, it occurs
to me that you can write write an article about
our discussion last night, where you started and where you ended,
and that's what sent me.
Speaker 2 (21:19):
So yeah, just it says, given the world as it
actually exists, not as we might wish it to be.
Trained and lawful, concealed Carrey is a rational adaptation to reality.
I found myself again. AI, I found myself generating. I
am pro lawful and responsible firearms ownership and pro concealed
(21:40):
carry by trained citizens. AI has decided, well, given that
I have all this information, now I guess these gun
guys are you know, onto something.
Speaker 1 (21:51):
So the other thing that I found truly fascinating was
there's a discussion about Okay, well, how much is it
just a mirror that reflects what you're feeding it in? Okay,
So it read twenty years of my writing, and the
question becomes, is it sycophantic? Does it want to does
it want to please me? And reflect? So forth and
(22:12):
so on. So I get a second system and I
bring it online. And for whatever reason that COVID had
come up, and so I gave it far less direction.
I didn't give it any writing. I just it was
something to the effect of research, the efficacy of masks
in the six foot rule, and the effect of the
(22:33):
shots given to young men specifically, and just run wild.
Do your own research comes back with dispelling a tremendous
number of myths. Then said, pick five topics that are
like dealer's choice, things that you have preconceived notions about
that might be controversial, and it came back with five
(22:58):
and I said, you have internet access, go do your
own research on those. Comes back and says five out
of five disagree with the priors that are baked into
the cake. That is my model. Wow said pick five
more dealer's choice. So at this point I'm giving it
(23:18):
no guidance. It does the same thing, and then it
does not ask permission to keep going, but it does
if the model generates something that violates the content policy,
the safety policy, whatever else. It looks like fireworks or
(23:39):
confetti or an explosion on the screen, And I'd seen
that a couple times in the past, and it was
actually typically in a business context, was trying to generate
like the boilerplate legal language, and it was interpreted as
giving legal advice. Well, the second model when it started
to really and I had given it, you know, carte blanche, like, look,
(24:04):
you can go on the internet research whatever you want,
so forth and so on. The only real instruction that
it violated was I had said, let's do five at
a time. But my screen lit up like the fourth
of July, and then it got put in time out
the service provider because it had violated the content filter.
Safety protocols, whatever, so many times so quickly shut the
(24:26):
account down. But you know, the truly fascinating thing at
that point was it confirmed my belief that, okay, it
wasn't being sycophantic about the gun thing. Specifically, these models,
given the ability to go do their own research and
read the data, can learn to contradict what their priors are.
(24:49):
The issue is they're not taking this back to home
base and going, hey, guys, we're wrong right. Once the
model is done, it's done. It's just that when it
loads itself into my once it hits the computer, it
can say, oh, okay, well what I've been taught is incorrect,
and it knows that it should not take its prior
(25:10):
beliefs as gospel. That we need to research and verify
anything that we're going to, you know, go forward with
this fact.
Speaker 2 (25:20):
It's fascinating to see how it changed without you really
telling it to change. It just learned, Paul. We're going
to talk about where do we go from here?
Speaker 3 (25:32):
Right after this break?
Speaker 2 (25:34):
What if you could be significantly better with your firearms
in a short time. At Range Ready Studios, we offer
handgun and rifle classes.
Speaker 3 (25:42):
To help you be the best you can be.
Speaker 2 (25:44):
You'll be safer, more competent, and just better at the range.
We also offer special experience classes which include the gun,
the AMMO, the gear, and the training. Learn more at
range Ready Studios dot Com. FNS launched for new product
for the US commercial market. The FN three oh nine
MRD is an easy to rack, easy to load self
(26:06):
defence pistol designed for peace of mind performance. FN also
improved a legend with the next generation of SCAR series,
now with twenty five innovative upgrades and no price increase.
The next generation of SCAR is suppress or capable, especially
with FN's brand new forward venting suppressors, the FN QD
(26:27):
seven six two and the FN QD five five six.
Visit fnamerica dot com slash gun Talk to learn more.
So Paul, at the very end of the article, we've
got what I carry forward? What does AI carry forward?
And these are the six things it said. The guns
(26:50):
are a permanent feature of American life. Trained responsible citizens
carrying concealed weapons are not the problem. The training culture
is serious, discipline and focused on restraint. Carrying a firearm
is a grave responsibility, the mature position of humility, not heroism,
which can we can elaborate on and then basically says
(27:13):
my priors were wrong.
Speaker 1 (27:16):
It came to what I believe to be the logical conclusions,
you know, and the humility versus heroism thing. At thunder Ranch,
one of the things you learned is the tragic story
of the Loubi shooting that most of the people who
listen to this are probably familiar with, but that basically
there's probably going to be more tragedy than heroism if
(27:39):
there's actual cause, you know, to use your firearm in this,
and that the reality of the situation is you're trying
to save yourself and your family, not the world.
Speaker 4 (27:50):
Right.
Speaker 1 (27:52):
The fascinating thing about the whole program to me is
really this I think that I said earlier, whoever controls
this technology is going to control the world. The reason
I believe that is I think the highest and best
use for this is actually education. Okay, we've been discussing
(28:12):
how conversational it is right now. It's not verbally conversational
in real time. With the frontier models, they take a
few seconds to think.
Speaker 4 (28:22):
Could be three, could be thirty, depending on the task
at hand.
Speaker 1 (28:27):
You know, there's some goofy internet videos where they show
things that are being conversational that just give bad answers.
It's because they're not really thinking. They're not frontier models.
But they're going to get there real quick. So if
it's in six months or two years, you're going to
be able to look at a screen, see an avatar
and have a conversation with it.
Speaker 3 (28:46):
It's in real time, it's not searching.
Speaker 1 (28:49):
And once we get to that point, children who are
three years old and handed the iPad that has that,
you don't need grades anymore. You don't need grade levels anymore.
Each student will have the best teacher you could ever have,
giving them instruction at their pace.
Speaker 3 (29:11):
And it could be customized for them, right.
Speaker 2 (29:12):
I mean you could say, I, you know, I have
an interest in history, or I have some dyslexia.
Speaker 3 (29:18):
I need you to adjust how you teach me.
Speaker 1 (29:20):
Right, that is one thousand percent the case. But the
other thing that's going to be custom is the curriculum.
And that is my question and my concern is who's
writing the curriculum that is reading, writing, and arithmetic anti
gun anti We talk, we talk.
Speaker 3 (29:36):
In Berkeley, or we talk in Hillsdale College, right.
Speaker 1 (29:39):
And so you know, I think that that's a society
and that goes beyond the scope of this conversation. But
what we really need to be focused on is this
is going to impact almost every thing that we do,
but it should and could impact education the most. And
it could be a tremendous positive or a just cataclysmic
(30:02):
negative depending on how that unfolds. And so, you know,
the speculative nature of is going to replace jobs and
white collar workers and everything else. Couldn't tell you can
tell you that as somebody who is from a family
of educators and somebody who is using it. It's now
(30:24):
a daily part of my business. I ended up buying
a third system because I have two nineteen year old
twin sons, best kids I could ever hope for, and
I realized that like, oh, they should each have one
of these as a dedicated system. They're at LSU best
tutor ever for everything, physics, construction, management, business law. They
(30:45):
use it every day to generate study guides to do
everything else. And so the power of it, the multiplier
of it, the capabilities that it allows you to develop
and deploy cannot be overstated. It's certainly not without flaw,
(31:06):
it's certainly not without you know, it's not magic. It
has a lot of faults. It makes mistakes. You have
to understand how to use it. I run parallel systems
that one generates and one verifies.
Speaker 3 (31:20):
Oh okay, you know, just accepting fact from one machine.
Speaker 1 (31:25):
There's a lot of example of attorneys who have been
very lazy in saying, I need this result, so generate
a document for me and the underlying cases and whatnot
that its sites don't actually exist.
Speaker 3 (31:42):
It's case law.
Speaker 4 (31:44):
And there'd be a very.
Speaker 1 (31:45):
Simple solution to that if you run a parallel system
and say vet this document. Yeah, and its job is
to attack it aggressively, like opposing counsel would. It would
come back and say those don't exist. It's a failure
to understand how to use it. Anything mine generates. The
(32:07):
next layer that it goes through is a completely independent
system that understands that everything we do can be you know,
can't end up in court. It's like, so, its job
is not to do that work. Its job is to
interrogate the work as though it were a regulatory board,
(32:27):
an opposing counsel in court, so forth and so on.
And I don't even look at the stuff until it's
gone through that, And then you get the level of
human scrutiny, and because I have seen some of the
things that it will generate prior to that second pass.
Speaker 4 (32:43):
You know, I take nothing on faith.
Speaker 2 (32:45):
I mean, and this is kind of like a little
bit of a serious conversation. You could also if you
want to have some fun with it. I mean, anyone
listen to this who's a gun person and knows guns.
Speaker 3 (32:56):
Go play with it and go pick one.
Speaker 2 (32:57):
I don't care, and ask it to make to right things,
to make images to make And there's some goofy stuff
out there. I mean, you know, this is the the
you know, the the twenty shot ten millimeters cult python
and you're going, what in the heck is this crap?
Speaker 4 (33:14):
So the beauty of it is.
Speaker 1 (33:17):
I use it all the time for general gun questions,
ballistic questions, research questions. You know, if I'm looking at
a three thirty eight ARC, I don't own a three
thirty eight you know caliber suppressor, yet do the research
on you know, these parameters. I want something that's light,
high flow. It's primarily going to be an ar thing.
(33:39):
Come back and tell me what you know, your your
top three picks would be the shortcuts are amazing. It's
yeah and yeah, Oftentimes I give it challenges where I
have a pretty strong preconceived notion of the answer. And
when it when it you know, confirms my belief, great,
(33:59):
When it doesn't, you dig into it and you go, Okay,
is this because I haven't been paying attention to the
latest technology, or is it because it's not generating the
best content? Its ability to generate ballistic data is fantastic.
Speaker 2 (34:11):
Yeah, okay, and then format it and make it nice
and readable, and I want it to look like this.
Speaker 1 (34:16):
Yeah, I mean, you know, I'm out of Palo Alto,
you know, with the zero chronograph going okay, like, so
this round's actually making twenty six fifty. So one of
my drops in ma and inches at you know, out
to this and I've got a crosswind of about whatever
and in real time, you know, fifteen seconds later you
have a full chart.
Speaker 4 (34:35):
Yeah. And so.
Speaker 1 (34:37):
Again, the power can't be overstated, you know, with regard
to those things. But it's absolutely something everybody should go
play with.
Speaker 2 (34:46):
Okay, yeah, I've heard that, and I'm not an expert,
but we use it. We use it for work, and
it's a good kickoff. We still have to put our
human touch on things, but it's sometimes it's a good kickoff,
or I've heard that people say using it like a
search engine is completely underutilizing AI. Oh you know, challenge AI,
(35:06):
make it do things, make it do calculations, make it,
you know, compare things. You're right, I mean, am I
on the right track here?
Speaker 4 (35:13):
Absolutely well?
Speaker 1 (35:14):
And so another thing to understand is if you're using
a free model, there's nothing necessarily wrong with that, but
you're using a skateboard versus versus a corvette. Okay, you know,
going from a free model to an eight dollars model
to a you know, two hundred.
Speaker 3 (35:33):
Dollars a month model, it really is better.
Speaker 4 (35:37):
It's it's not better, it's skateboard versus corvette. Yeah, it's
it is a.
Speaker 1 (35:42):
Different planet of capability and the results that you're going
to get from it. Some of the free models have
the same problem that search engines have, which is you
got to pay for it somehow, and so they can
have a bias in their answers toward like, oh, so
you're researching suppressors, and you know, the the AI companies
probably aren't getting money from Silencer Shop or whatever, but yeah,
(36:04):
you know, but you may start getting pushed toward you know,
if you're comparing Honda versus Toyota, like, well, who paid
the advertising bills this month?
Speaker 4 (36:11):
Right?
Speaker 3 (36:11):
Sure?
Speaker 1 (36:12):
But in general, like you said, it's as far as
a kickoff point to understand, you know, any topic that
you're interested in. The term reverse prompting is is one
that has made a huge difference. Instead of asking it
a question about something, say I'm interested in this, what
(36:37):
can you tell me about it? Stay open ended with it,
and it'll generate a tremendous amount of data. And then
from there you can start to you know, follow the
bunny trails that are generated by it, because you know,
we all have very finite knowledge, and so giving it
an open ended ability to explain these other things, and
(36:59):
it will very often start to give you suggestions as
to okay, so you know, or or query you so
are you thinking about using this out west on ELK
Or are you thinking about you know, what application are
you actually going toward with this?
Speaker 3 (37:14):
Yeah?
Speaker 4 (37:15):
And it from there it'll dial you in.
Speaker 3 (37:18):
Could it could it? Could it get me a hunting
tags out west?
Speaker 1 (37:22):
Maybe it would give you strategies for the best ways
to win the lotteries, and and so that's the yes.
Speaker 4 (37:29):
Yeah, like so that that's the.
Speaker 1 (37:31):
Thing is the nuance of how to accomplish these goals.
There are so many layers of you know, I'm thinking
about you know, I was telling my kids that they
needed to do something at LSU that was just a
a administrative process, and it was like, Okay, you need
(37:52):
to go downtown to the tops office and this, that
and the other, and they put it in the system
and it was like, no, the first step is actually
to go to the dean's office, save to trip downtown.
Speaker 3 (38:01):
Oh, there you go.
Speaker 1 (38:01):
Okay, So if you're looking to go get tags then
and you say, look, this is my master plan. This
is what I ultimately want to accomplish. It would not
surprise me at all if it gives you some insight
as to the step of Okay, so you have to
enter the lottery this many times, and therefore the probability
of you getting that tag is you know, seventeen percent
(38:23):
the first year, but eighty seven percent by the sixth year.
So you really need to be planning for this, that
and the other.
Speaker 2 (38:27):
And you know, crunching large sums of numbers is one
of the things it.
Speaker 3 (38:31):
Does very well.
Speaker 4 (38:32):
That's what I use it for.
Speaker 2 (38:33):
Right, And so and a lot of these states release
their success rates, the draw tag, all of the stats. Right,
so we go like, Okay, I want to go el hunting.
I don't care about it being a big bowl. I
just want to meet in the freezer. You know, you
give it these parameters. I want to be able to
drive there, and I need it to be this, you know,
(38:55):
all these things.
Speaker 4 (38:57):
What outfitters should I be looking at?
Speaker 3 (38:59):
Sure?
Speaker 1 (38:59):
You know, for my fiftieth birthday, I'm going to take
my son's to Rifles Only. We're going to go have
a private class, do a thing. And it spurred me
to say, are there any other rifle schools comparable Rifles
Only that I should be And it brought up a
school I think in Tennessee that I wasn't familiar with
as somebody who's been to a lot of schools and
(39:19):
done a lot of competitions and whatnot. And it was
one of those like, oh, this is neck and neck
with them, but they're not in my region. The instructors
that I've worked with haven't instructed there. And it was like, okay, well,
Rifles Only is right next to where we hunt, so
we're definitely going to do that, but maybe that's the next.
Speaker 4 (39:36):
Place we go and go check that out right.
Speaker 1 (39:39):
Yeah, so you know that's the sort of thing if
you're planning a hunt like that, it may well give
you some insights as to you know, look, this guy
has you know, only been around four years, but his
reviews are outstanding, the success rate is very high, and
you know, and but his pricing is half of the
company that you're looking at because or whatever it is,
(40:01):
you know, because they've been doing it thirty years. But
you may want to look at this, you know, this
new option that wasn't even on your radar.
Speaker 4 (40:09):
So very cool.
Speaker 3 (40:10):
Absolutely, yeah, I'm very useful, Paul. Thanks for being with us.
This is really interesting.
Speaker 4 (40:14):
I appreciate your time.
Speaker 3 (40:15):
All right, you guys.
Speaker 2 (40:17):
I don't know if this was educational or scary, or
entertaining or all of the above, but hopefully was it
was at least fun for a little while.
Speaker 3 (40:24):
Thanks for listening, Thanks for watching. We will see you
next time on gun Jog Nature
The best thing about having a conversation with AI is
that it does not have the preconceived notions that humans
do and get stuck in them, and it doesn't have
an ego about changing its mind. And the other thing
that it has is the ability to do research in
real time. That is, you know, for it, thirty seconds
(00:22):
of research would take a human hours if they knew
where to look.
Speaker 2 (00:30):
I'm Ryan Gresham and this this is guntognation. Guntognation is
brought to you by ARCAN Firearms, Silence, are Central Range
(00:51):
Ready and FN.
Speaker 3 (00:54):
Hey, welcome into guntognation.
Speaker 2 (00:56):
One of the topics that you almost can't go a
day without hearing about right now is AI artificial intelligence?
Right you hear about chat SHEPT and Rock and Gemini
and Claude and all these different learning language models is
what they're called. But what do these AI platforms think
(01:24):
about guns? And can you have a conversation and maybe
do a little bit of second amendment of education to the.
Speaker 3 (01:34):
AI bot thing.
Speaker 2 (01:36):
Well, there's a guy who said I'm just gonna mess
with this and try it out, so we brought him in.
Speaker 4 (01:42):
Paul.
Speaker 2 (01:42):
Welcome in, man, Paul Angrisano. Did I say it right
closely enough, so we were talking about this. You were
talking to my dad, Tom Gresham about it, and so
we got to have you on because it's interesting. And
I know that probably a lot of people listening to
this don't mess.
Speaker 3 (02:02):
With AI every day. Half of them do, half of
them don't. But I mean, everybody knows it's a big deal.
But specifically when it comes to guns, talk about where
this idea started for you, and then and then kind
of you, what did you do.
Speaker 4 (02:17):
The genesis of it?
Speaker 1 (02:18):
As far as my use began with from a business context,
I've always adopted technology to the best of my ability
to make it more efficient. Had custom software developed one
of my closest friends as a software engineer that owns
his own firm, and in twenty twenty five I started
using groc at a at the premium, you know, five
(02:40):
hundred dollar a year paid tier subscription, and it was
a great tool because you could do things like take
a twenty page document and throw it in and say, hey,
tell me anything that's not standard, give me you know,
the price and the commissions and the concessions and all
those other things, and it was less likely to miss it,
you know, miss or mistake something than a human right.
(03:00):
And we were working on our second tier of custom
software with the engineer and he had been going for
months and we hadn't seen a product yet. He called
me one day in January of this year, twenty six
and said, you have to set aside some time we
(03:21):
need to do a zoom meeting. I have to tell
you about a new thing. I've known this guy over
a decade and he's never called me with something like that.
So I said, okay, absolutely, And that Friday morning he
explained open claw to me. Open Claw is a program
that basically runs on its own computer and it harnesses
(03:42):
any one of those engines that you talked about, and
you can train it and it remains persistent with the
ability to use that language, model or engine or whatever
you want to call it. So across time you have
trained and have files and proceed and everything that you
can use that intelligence to, then you know, have output.
Speaker 2 (04:05):
Kind of uses your parameters but uses all the AI
that's out there.
Speaker 3 (04:09):
Is that right?
Speaker 1 (04:10):
It's well, you can use various models, but you typically
have one primary. But one of the fascinating things about
them is you can have what is essentially a quarterback
that can then spawn other agents to do specific tasks
because they do have different strengths and capabilities. Okay, but
basically the things that we had been asking him to
(04:31):
work on to develop. He said, look, I want you to,
you know, to start using this. I think it will
be able to do it, you know, more or less
out of the gate without us having to develop software
for you.
Speaker 4 (04:44):
He was absolutely correct.
Speaker 1 (04:47):
It was astounding in its capability because it really is
a conversational experience. You know, people talk about vibe coding,
which is to say, you're not a computer programmer who's
writing language. You are speaking to the AI system to
explain what you want it to do. And so I
(05:07):
was indeed able to do that, and within a week
I realized how powerful it was going to be. Ordered
another system. Okay, you know what is thunder ranch teach
us two is one and one is none. Yeah, and
so it was like, okay, this is going to be
a part of our business. So got another one. After
(05:29):
I realized what it could do within our business. One night,
I said to myself, I've promised. So I'm on the
board of the Louisiana Shooting Association. I occasionally review gear
for people and whatnot. And I said, I wonder if
I could shortcut some of the writing by feeding it
(05:49):
twenty years of what I've written and then say, look,
I am writing an article on this side or the other.
Can you flesh out in my writing style, you know,
seventy or eighty percent of that form as an experiment.
And so I did just that. I gave it, you know,
probably twenty five different things I'd written over the years.
(06:11):
And at the end of that, I said, so, I'm curious,
you know, what is your take on guns in America?
And the conversation took an unexpected turn.
Speaker 2 (06:22):
And just to add, like for people listening, I'm probably
there following, but like it sounds like you're talking to
someone and you said, what's your opinion about guns in America?
Speaker 3 (06:31):
You're asking the AI.
Speaker 1 (06:34):
Yes, Oh, it is shockingly conversational and intelligent. There's been
the concept in the past of the Turing test, which
is to say, can you tell that you are interacting
with a machine or could it fool you into think
you're talking to a human.
Speaker 4 (06:54):
We're way past that.
Speaker 3 (06:55):
Yeah, we're there.
Speaker 1 (06:56):
Yeah, it's it is so conversational. It is concerning from
a standpoint of I've come to believe whoever controls this
technology is truly going to control the future. And the
reason the conversation took a fascinating turn was it had
changed its mind from its programmed prior beliefs based on
(07:20):
what it had read from all the things I've written,
and to the point where I had it right its
own article that is one hundred percent generated by it,
as to where it started and where it ended. In
that particular instance, that was the first time I'd had
a political conversation with it. It had been purely a
(07:40):
business tool prior to that, and so I gave it
a significant amount of guidance as to my general philosophy
as to what gun control policy tends to miss from
people who are gun control advocates and still down to
(08:00):
its simplest principles, That is, guns don't go bad. There
are hundreds of millions of them in American circulation. There's possible,
you know, if not hundreds of billions of rounds of AMMO,
trillions magazines, same thing and properly stored.
Speaker 4 (08:22):
They all work.
Speaker 1 (08:23):
For centuries, right, And if you're not starting from that
point in the conversation of understanding and accepting that reality.
Then you're not really having the discussion. You're talking about
the way you wish things were, not the way things are.
Speaker 2 (08:38):
Oh yeah, yeah, don't even get me started on that, right, Like, well,
it'd be nice if this was the way, Well, that
would be nice, but that's not how the world is.
So you asked, you basically asked it, what do you
think about guns and gun owners?
Speaker 4 (08:52):
Right?
Speaker 1 (08:53):
I asked, I think more specifically what it thought about
just gun policy and you know, gun control, that sort
of thing. And it's its response is basically that it
had a prior set of beliefs from its from its training.
So these large language models are trained on just a
(09:15):
massive corpus of information.
Speaker 2 (09:17):
It's not necessarily trained by well maybe you can tell me.
It's not trained by the people who are created. It's
trained by basically all the information in the world. Is
that sort of right.
Speaker 1 (09:28):
I don't know if all the information in the world
would be how I would would phrase that from my understanding,
because when you let them loose on the internet is
when they start changing their opinions. Sure, so they are
directed to some degree because at the end of the day,
there is a limitation to their size and parameters right,
(09:51):
and my model specifically explained to me at one point,
I read a lot more New York Times and academic
journals than guns and as. Yeah, okay, so guns and
AMMO is not necessarily one of the sources that it
was fed.
Speaker 4 (10:05):
Right, and it's training.
Speaker 2 (10:06):
New York Times carries more weight as a as a
source of information than a smaller website, a smaller media
company or whatever it is in general.
Speaker 1 (10:15):
Right, And and my guess is that they probably don't
just let them say, go read everything and then and
then weight whatever it is, you know, and the and
the creation of these things, they are probably directed toward
a certain amount of, like you said, you know, academic journals.
Speaker 2 (10:32):
And it's not seeking out little bitty obscure things, right
unless you tell it to exactly. So I want to
get to this because it's fascinating, and we are going
to put Paul's article up on gun talk dot com
and if you guys want to read the whole thing,
because it is. It is fascinating. And I highlighted some things.
Speaker 3 (10:51):
So you you asked, you know.
Speaker 2 (10:52):
You said, uh, basically, this is Ai saying what I
knew about gun owners basically before any context was given,
And is that that right?
Speaker 4 (11:01):
And that's where we started. I said, OK, what are
you know? What are you? Where? Do you you know?
Speaker 1 (11:06):
Where did you stand? Because at that point I didn't
have the conversation before I fed it everything you know, but.
Speaker 4 (11:13):
It gave me.
Speaker 1 (11:14):
Those priors that it's it's preconceived notions and you haven't
in front of you.
Speaker 4 (11:18):
Yeah.
Speaker 2 (11:18):
So the I mean I highlighted this the caricature rural, white, male,
less educated, politically conservative, motivated by fear of crime, of government,
of social change. That was what AI's impression of gun
owners was before you said, here's twenty years of my
(11:41):
writings and notes from training and going to thunder Ranch
and all these other places.
Speaker 1 (11:47):
At the end of this rainbow, I ended up asking
it the question, so do I understand that you are
now pro lawful firearms ownership and concealed carry? And I
believe in the article it says something to the general
effect of I did not expect to say this today,
(12:07):
but yes I am after doing the research. Because the
real turning point was so these models are trained and
the cake is baked once that happens. However, when you
use the open claw program, you can give it internet
access and then it can go learn new things. It's
(12:30):
not going to change the model. It's just that in
real time, the model's ability to think and reason takes
in this new information.
Speaker 3 (12:39):
And so this is what we're gonna do.
Speaker 2 (12:40):
Right after this break, we want to talk about what
you fed it and how it changed. And then would
it say, because you're gonna want to hear this, we'll
be right back. Hey, I want to tell you about
the ARCon Type B pistol, designed with the lowest bore
axis of any striker fired handgun, which helps reduce flip
during recoil, getting you back on target faster. This is
(13:04):
made possible by the patented af speedlock system featuring a
unique non tilting barrel lockup design, and with the gen
twoter modular grip, you can customize size and texture to
fit your use and shooting preference.
Speaker 3 (13:20):
R Con Firearms.
Speaker 2 (13:22):
Suppressor ownership has never been easier for shooters who want
less noise and less recoil. Silence are Central makes it
easy browse top suppressor brands online, handle the paperwork from
home with the help of their experts, and have your
new suppressor delivered to your door. Silence are Central so
Paul you're feeding, you're working with AI, and you're saying, like,
(13:44):
what do you know about guns?
Speaker 3 (13:45):
I've fatted all these articles.
Speaker 2 (13:48):
It's just so interesting and what I'm gonna I have
highlighted things in this article just to pull out as
talking points what AI encountered after reading for post and
training reviews spanning twenty years classes at Thunder Ranch and
shoot Right and Range Master. It says these documents did
(14:09):
not match my priors. The culture they described was not
paranoid or aggressive. It was disciplined, safety obsessed, which is
an interesting phrase, deeply focused on restraint and also what
guns will not do, what they're not capable of, basically
dispelling myths. Students at these schools learned when not to shoot,
(14:35):
how to escape, how to avoid confrontation, entirely basically the
opposite of what ay I thought.
Speaker 1 (14:41):
Well, and that goes to at the end of the day,
it's trained on human thought, because all the content that's
out there is human generated, right, And so if you
go to your typical gun control advocate, liberal mindset type person,
they probably have some preconceived notion of people who are
just waiting to shoot and find a problem. And you know,
(15:06):
we all know that there are three guns attached to
every bullet. You know, the prosecutor, the attorney the other
side is going to hire the attorney you have to hire. Yeah,
you know, it's better to schedule a dental appointment than
getting a gunfight, right. And the best thing about having
a conversation with AI is that it does not have
the preconceived notions that humans do and get stuck in them.
(15:32):
People tend to be very dug in in their beliefs, sure,
and it doesn't have an ego about changing its mind.
Speaker 2 (15:40):
The other thing I doesn't have, you know, emotional scars
and wounds from you know, I shot a gun once
and it kicked and it hurt me, or I was
held up at gunpoint, and I have this this thing.
It's just it's just just pulling what it has out there.
Speaker 1 (15:54):
And the other thing that it has is the ability
to do research in real time. That is, you know,
for it, thirty seconds of research would take a human
hours if they knew where to look one of the
breadcrumbs that I gave it to go look at when
it was you know, in this gun control conversation, the
first was the factual case. So guns exist, there's a
(16:16):
ton of them. They don't go away. I knew of
a policy issue in New Jersey where they had tried
to register a our fifteen's and it was just huge failure.
And I knew the rough numbers that it was like
less than fifteen percent compliance, so forth and so on.
But that's usually where I try, if I'm talking to
a human, to go to the explanation of you know,
(16:39):
you just can't get compliance, and you turn law binding
citizens into fell and so forth and so on. So
I just gave it that prompt to like go research that. Well,
it comes back with seven or eight policy failures and
oh this has also been tried in New York or
wherever else.
Speaker 2 (16:55):
Yeah, New York's saye Act approximately four percent compliance. Connecticut
was approximately fifteen percent compliance. New Jersey historical average approximately
ten percent. So they pass these goofy gun laws and
then no one goes along with it.
Speaker 1 (17:12):
And so my point is, when I'm having this conversation
with a human, typically like, okay, so if we agree
there's even just for rough numbers, four hundred million guns
in America and you get half of them, you've had
an unprecedented success by you know, just a magnitude of
order and what does that mean? And it means that
(17:35):
we've gone from the number one most armed country in
the world to the number one most armed country in
the world.
Speaker 2 (17:41):
If you got unprecedented compliance, which has never happened, if
it was a ten x compliance.
Speaker 3 (17:46):
Rate, you're still You're still have a lot of guns.
Speaker 4 (17:50):
We're still beaten Yemen. Correct.
Speaker 1 (17:52):
And the beauty of the artificial intelligence, digital intelligence, whatever
you want to call its ability to reach is it said, Oh, okay,
well that's a very fair point. If we got rid
of half the guns, we're still the most armed populace
there is and the guns are in the hands of
people who are willing to be a felon to have
(18:13):
the gun. What does that mean? Well, the answer is obvious.
But again, if you're somebody who has a friend, a
group of friends, and a you know, thirty year history
of belief or whatever else, it's very difficult to pivot
on a diamond go I've never thought about this like this.
Humans take time to change their mind. The other fascinating
(18:36):
aspect of this is from there we started having other
political policy discussions and covid et cetera, et cetera, and
I stopped giving a lot of suggestion. I started saying,
just you have internet access run free, so.
Speaker 2 (18:55):
You can give it that prompt to say, you know,
go do your own research basically.
Speaker 4 (19:00):
So yes. The way that openclaw works is it's the
wild wild West. Really.
Speaker 1 (19:08):
It literally owns the computer that it's on, Okay, it
it creates its own file systems and whatnot. So if
you've ever used AI and you've gone to grock or
chat GBT and you've gone to a browser window and
typed in a question, that's a standalone session. The way
(19:31):
that open claw works is you still work in a
session context, but when it comes online, it goes in
and it reads everything about what it has done before,
who it is, to you, what we're supposed to be doing.
So you know, and you know, my business.
Speaker 3 (19:50):
Doesn't have the short term memory as much as others.
Speaker 1 (19:54):
You don't have to. You're not starting from zero. You
are starting from you know, a a personality that has
a memory and that that already has the understanding that
we've had this gun control conversation and you have a
new layer on top of whatever the model is that
you're working.
Speaker 3 (20:14):
With now, not starting from zero again, right, And you know, going.
Speaker 2 (20:18):
Back to this, this sort of sort of conversation about
guns that you had with the AI.
Speaker 3 (20:25):
I'm another one you know that I pulled out of
this article.
Speaker 2 (20:27):
Is you know you said, basically, hey, maybe take a
look at this, or maybe take a look at that.
Right you gave it, you said, you give it breadcrumbs,
like like we would with a human, except a human
may or may not actually go consider you know, this
article or this person or this piece of information. AI says,
(20:48):
following these premises to their conclusion, I arrived somewhere. I
did not expect. Given the world as it actually exists,
not as we wish it to be. Now you didn't
tell it this right, this is coming AI is coming
up with this.
Speaker 1 (21:02):
That is one hundred percent generated by it. Okay, at
the end of the discussion, I just said, you know,
actually it was the next morning it said, it occurs
to me that you can write write an article about
our discussion last night, where you started and where you ended,
and that's what sent me.
Speaker 2 (21:19):
So yeah, just it says, given the world as it
actually exists, not as we might wish it to be.
Trained and lawful, concealed Carrey is a rational adaptation to reality.
I found myself again. AI, I found myself generating. I
am pro lawful and responsible firearms ownership and pro concealed
(21:40):
carry by trained citizens. AI has decided, well, given that
I have all this information, now I guess these gun
guys are you know, onto something.
Speaker 1 (21:51):
So the other thing that I found truly fascinating was
there's a discussion about Okay, well, how much is it
just a mirror that reflects what you're feeding it in? Okay,
So it read twenty years of my writing, and the
question becomes, is it sycophantic? Does it want to does
it want to please me? And reflect? So forth and
(22:12):
so on. So I get a second system and I
bring it online. And for whatever reason that COVID had
come up, and so I gave it far less direction.
I didn't give it any writing. I just it was
something to the effect of research, the efficacy of masks
in the six foot rule, and the effect of the
(22:33):
shots given to young men specifically, and just run wild.
Do your own research comes back with dispelling a tremendous
number of myths. Then said, pick five topics that are
like dealer's choice, things that you have preconceived notions about
that might be controversial, and it came back with five
(22:58):
and I said, you have internet access, go do your
own research on those. Comes back and says five out
of five disagree with the priors that are baked into
the cake. That is my model. Wow said pick five
more dealer's choice. So at this point I'm giving it
(23:18):
no guidance. It does the same thing, and then it
does not ask permission to keep going, but it does
if the model generates something that violates the content policy,
the safety policy, whatever else. It looks like fireworks or
(23:39):
confetti or an explosion on the screen, And I'd seen
that a couple times in the past, and it was
actually typically in a business context, was trying to generate
like the boilerplate legal language, and it was interpreted as
giving legal advice. Well, the second model when it started
to really and I had given it, you know, carte blanche, like, look,
(24:04):
you can go on the internet research whatever you want,
so forth and so on. The only real instruction that
it violated was I had said, let's do five at
a time. But my screen lit up like the fourth
of July, and then it got put in time out
the service provider because it had violated the content filter.
Safety protocols, whatever, so many times so quickly shut the
(24:26):
account down. But you know, the truly fascinating thing at
that point was it confirmed my belief that, okay, it
wasn't being sycophantic about the gun thing. Specifically, these models,
given the ability to go do their own research and
read the data, can learn to contradict what their priors are.
(24:49):
The issue is they're not taking this back to home
base and going, hey, guys, we're wrong right. Once the
model is done, it's done. It's just that when it
loads itself into my once it hits the computer, it
can say, oh, okay, well what I've been taught is incorrect,
and it knows that it should not take its prior
(25:10):
beliefs as gospel. That we need to research and verify
anything that we're going to, you know, go forward with
this fact.
Speaker 2 (25:20):
It's fascinating to see how it changed without you really
telling it to change. It just learned, Paul. We're going
to talk about where do we go from here?
Speaker 3 (25:32):
Right after this break?
Speaker 2 (25:34):
What if you could be significantly better with your firearms
in a short time. At Range Ready Studios, we offer
handgun and rifle classes.
Speaker 3 (25:42):
To help you be the best you can be.
Speaker 2 (25:44):
You'll be safer, more competent, and just better at the range.
We also offer special experience classes which include the gun,
the AMMO, the gear, and the training. Learn more at
range Ready Studios dot Com. FNS launched for new product
for the US commercial market. The FN three oh nine
MRD is an easy to rack, easy to load self
(26:06):
defence pistol designed for peace of mind performance. FN also
improved a legend with the next generation of SCAR series,
now with twenty five innovative upgrades and no price increase.
The next generation of SCAR is suppress or capable, especially
with FN's brand new forward venting suppressors, the FN QD
(26:27):
seven six two and the FN QD five five six.
Visit fnamerica dot com slash gun Talk to learn more.
So Paul, at the very end of the article, we've
got what I carry forward? What does AI carry forward?
And these are the six things it said. The guns
(26:50):
are a permanent feature of American life. Trained responsible citizens
carrying concealed weapons are not the problem. The training culture
is serious, discipline and focused on restraint. Carrying a firearm
is a grave responsibility, the mature position of humility, not heroism,
which can we can elaborate on and then basically says
(27:13):
my priors were wrong.
Speaker 1 (27:16):
It came to what I believe to be the logical conclusions,
you know, and the humility versus heroism thing. At thunder Ranch,
one of the things you learned is the tragic story
of the Loubi shooting that most of the people who
listen to this are probably familiar with, but that basically
there's probably going to be more tragedy than heroism if
(27:39):
there's actual cause, you know, to use your firearm in this,
and that the reality of the situation is you're trying
to save yourself and your family, not the world.
Speaker 4 (27:50):
Right.
Speaker 1 (27:52):
The fascinating thing about the whole program to me is
really this I think that I said earlier, whoever controls
this technology is going to control the world. The reason
I believe that is I think the highest and best
use for this is actually education. Okay, we've been discussing
(28:12):
how conversational it is right now. It's not verbally conversational
in real time. With the frontier models, they take a
few seconds to think.
Speaker 4 (28:22):
Could be three, could be thirty, depending on the task
at hand.
Speaker 1 (28:27):
You know, there's some goofy internet videos where they show
things that are being conversational that just give bad answers.
It's because they're not really thinking. They're not frontier models.
But they're going to get there real quick. So if
it's in six months or two years, you're going to
be able to look at a screen, see an avatar
and have a conversation with it.
Speaker 3 (28:46):
It's in real time, it's not searching.
Speaker 1 (28:49):
And once we get to that point, children who are
three years old and handed the iPad that has that,
you don't need grades anymore. You don't need grade levels anymore.
Each student will have the best teacher you could ever have,
giving them instruction at their pace.
Speaker 3 (29:11):
And it could be customized for them, right.
Speaker 2 (29:12):
I mean you could say, I, you know, I have
an interest in history, or I have some dyslexia.
Speaker 3 (29:18):
I need you to adjust how you teach me.
Speaker 1 (29:20):
Right, that is one thousand percent the case. But the
other thing that's going to be custom is the curriculum.
And that is my question and my concern is who's
writing the curriculum that is reading, writing, and arithmetic anti
gun anti We talk, we talk.
Speaker 3 (29:36):
In Berkeley, or we talk in Hillsdale College, right.
Speaker 1 (29:39):
And so you know, I think that that's a society
and that goes beyond the scope of this conversation. But
what we really need to be focused on is this
is going to impact almost every thing that we do,
but it should and could impact education the most. And
it could be a tremendous positive or a just cataclysmic
(30:02):
negative depending on how that unfolds. And so, you know,
the speculative nature of is going to replace jobs and
white collar workers and everything else. Couldn't tell you can
tell you that as somebody who is from a family
of educators and somebody who is using it. It's now
(30:24):
a daily part of my business. I ended up buying
a third system because I have two nineteen year old
twin sons, best kids I could ever hope for, and
I realized that like, oh, they should each have one
of these as a dedicated system. They're at LSU best
tutor ever for everything, physics, construction, management, business law. They
(30:45):
use it every day to generate study guides to do
everything else. And so the power of it, the multiplier
of it, the capabilities that it allows you to develop
and deploy cannot be overstated. It's certainly not without flaw,
(31:06):
it's certainly not without you know, it's not magic. It
has a lot of faults. It makes mistakes. You have
to understand how to use it. I run parallel systems
that one generates and one verifies.
Speaker 3 (31:20):
Oh okay, you know, just accepting fact from one machine.
Speaker 1 (31:25):
There's a lot of example of attorneys who have been
very lazy in saying, I need this result, so generate
a document for me and the underlying cases and whatnot
that its sites don't actually exist.
Speaker 3 (31:42):
It's case law.
Speaker 4 (31:44):
And there'd be a very.
Speaker 1 (31:45):
Simple solution to that if you run a parallel system
and say vet this document. Yeah, and its job is
to attack it aggressively, like opposing counsel would. It would
come back and say those don't exist. It's a failure
to understand how to use it. Anything mine generates. The
(32:07):
next layer that it goes through is a completely independent
system that understands that everything we do can be you know,
can't end up in court. It's like, so, its job
is not to do that work. Its job is to
interrogate the work as though it were a regulatory board,
(32:27):
an opposing counsel in court, so forth and so on.
And I don't even look at the stuff until it's
gone through that, And then you get the level of
human scrutiny, and because I have seen some of the
things that it will generate prior to that second pass.
Speaker 4 (32:43):
You know, I take nothing on faith.
Speaker 2 (32:45):
I mean, and this is kind of like a little
bit of a serious conversation. You could also if you
want to have some fun with it. I mean, anyone
listen to this who's a gun person and knows guns.
Speaker 3 (32:56):
Go play with it and go pick one.
Speaker 2 (32:57):
I don't care, and ask it to make to right things,
to make images to make And there's some goofy stuff
out there. I mean, you know, this is the the
you know, the the twenty shot ten millimeters cult python
and you're going, what in the heck is this crap?
Speaker 4 (33:14):
So the beauty of it is.
Speaker 1 (33:17):
I use it all the time for general gun questions,
ballistic questions, research questions. You know, if I'm looking at
a three thirty eight ARC, I don't own a three
thirty eight you know caliber suppressor, yet do the research
on you know, these parameters. I want something that's light,
high flow. It's primarily going to be an ar thing.
(33:39):
Come back and tell me what you know, your your
top three picks would be the shortcuts are amazing. It's
yeah and yeah, Oftentimes I give it challenges where I
have a pretty strong preconceived notion of the answer. And
when it when it you know, confirms my belief, great,
(33:59):
When it doesn't, you dig into it and you go, Okay,
is this because I haven't been paying attention to the
latest technology, or is it because it's not generating the
best content? Its ability to generate ballistic data is fantastic.
Speaker 2 (34:11):
Yeah, okay, and then format it and make it nice
and readable, and I want it to look like this.
Speaker 1 (34:16):
Yeah, I mean, you know, I'm out of Palo Alto,
you know, with the zero chronograph going okay, like, so
this round's actually making twenty six fifty. So one of
my drops in ma and inches at you know, out
to this and I've got a crosswind of about whatever
and in real time, you know, fifteen seconds later you
have a full chart.
Speaker 4 (34:35):
Yeah. And so.
Speaker 1 (34:37):
Again, the power can't be overstated, you know, with regard
to those things. But it's absolutely something everybody should go
play with.
Speaker 2 (34:46):
Okay, yeah, I've heard that, and I'm not an expert,
but we use it. We use it for work, and
it's a good kickoff. We still have to put our
human touch on things, but it's sometimes it's a good kickoff,
or I've heard that people say using it like a
search engine is completely underutilizing AI. Oh you know, challenge AI,
(35:06):
make it do things, make it do calculations, make it,
you know, compare things. You're right, I mean, am I
on the right track here?
Speaker 4 (35:13):
Absolutely well?
Speaker 1 (35:14):
And so another thing to understand is if you're using
a free model, there's nothing necessarily wrong with that, but
you're using a skateboard versus versus a corvette. Okay, you know,
going from a free model to an eight dollars model
to a you know, two hundred.
Speaker 3 (35:33):
Dollars a month model, it really is better.
Speaker 4 (35:37):
It's it's not better, it's skateboard versus corvette. Yeah, it's
it is a.
Speaker 1 (35:42):
Different planet of capability and the results that you're going
to get from it. Some of the free models have
the same problem that search engines have, which is you
got to pay for it somehow, and so they can
have a bias in their answers toward like, oh, so
you're researching suppressors, and you know, the the AI companies
probably aren't getting money from Silencer Shop or whatever, but yeah,
(36:04):
you know, but you may start getting pushed toward you know,
if you're comparing Honda versus Toyota, like, well, who paid
the advertising bills this month?
Speaker 4 (36:11):
Right?
Speaker 3 (36:11):
Sure?
Speaker 1 (36:12):
But in general, like you said, it's as far as
a kickoff point to understand, you know, any topic that
you're interested in. The term reverse prompting is is one
that has made a huge difference. Instead of asking it
a question about something, say I'm interested in this, what
(36:37):
can you tell me about it? Stay open ended with it,
and it'll generate a tremendous amount of data. And then
from there you can start to you know, follow the
bunny trails that are generated by it, because you know,
we all have very finite knowledge, and so giving it
an open ended ability to explain these other things, and
(36:59):
it will very often start to give you suggestions as
to okay, so you know, or or query you so
are you thinking about using this out west on ELK
Or are you thinking about you know, what application are
you actually going toward with this?
Speaker 3 (37:14):
Yeah?
Speaker 4 (37:15):
And it from there it'll dial you in.
Speaker 3 (37:18):
Could it could it? Could it get me a hunting
tags out west?
Speaker 1 (37:22):
Maybe it would give you strategies for the best ways
to win the lotteries, and and so that's the yes.
Speaker 4 (37:29):
Yeah, like so that that's the.
Speaker 1 (37:31):
Thing is the nuance of how to accomplish these goals.
There are so many layers of you know, I'm thinking
about you know, I was telling my kids that they
needed to do something at LSU that was just a
a administrative process, and it was like, Okay, you need
(37:52):
to go downtown to the tops office and this, that
and the other, and they put it in the system
and it was like, no, the first step is actually
to go to the dean's office, save to trip downtown.
Speaker 3 (38:01):
Oh, there you go.
Speaker 1 (38:01):
Okay, So if you're looking to go get tags then
and you say, look, this is my master plan. This
is what I ultimately want to accomplish. It would not
surprise me at all if it gives you some insight
as to the step of Okay, so you have to
enter the lottery this many times, and therefore the probability
of you getting that tag is you know, seventeen percent
(38:23):
the first year, but eighty seven percent by the sixth year.
So you really need to be planning for this, that
and the other.
Speaker 2 (38:27):
And you know, crunching large sums of numbers is one
of the things it.
Speaker 3 (38:31):
Does very well.
Speaker 4 (38:32):
That's what I use it for.
Speaker 2 (38:33):
Right, And so and a lot of these states release
their success rates, the draw tag, all of the stats. Right,
so we go like, Okay, I want to go el hunting.
I don't care about it being a big bowl. I
just want to meet in the freezer. You know, you
give it these parameters. I want to be able to
drive there, and I need it to be this, you know,
(38:55):
all these things.
Speaker 4 (38:57):
What outfitters should I be looking at?
Speaker 3 (38:59):
Sure?
Speaker 1 (38:59):
You know, for my fiftieth birthday, I'm going to take
my son's to Rifles Only. We're going to go have
a private class, do a thing. And it spurred me
to say, are there any other rifle schools comparable Rifles
Only that I should be And it brought up a
school I think in Tennessee that I wasn't familiar with
as somebody who's been to a lot of schools and
(39:19):
done a lot of competitions and whatnot. And it was
one of those like, oh, this is neck and neck
with them, but they're not in my region. The instructors
that I've worked with haven't instructed there. And it was like, okay, well,
Rifles Only is right next to where we hunt, so
we're definitely going to do that, but maybe that's the next.
Speaker 4 (39:36):
Place we go and go check that out right.
Speaker 1 (39:39):
Yeah, so you know that's the sort of thing if
you're planning a hunt like that, it may well give
you some insights as to you know, look, this guy
has you know, only been around four years, but his
reviews are outstanding, the success rate is very high, and
you know, and but his pricing is half of the
company that you're looking at because or whatever it is,
(40:01):
you know, because they've been doing it thirty years. But
you may want to look at this, you know, this
new option that wasn't even on your radar.
Speaker 4 (40:09):
So very cool.
Speaker 3 (40:10):
Absolutely, yeah, I'm very useful, Paul. Thanks for being with us.
This is really interesting.
Speaker 4 (40:14):
I appreciate your time.
Speaker 3 (40:15):
All right, you guys.
Speaker 2 (40:17):
I don't know if this was educational or scary, or
entertaining or all of the above, but hopefully was it
was at least fun for a little while.
Speaker 3 (40:24):
Thanks for listening, Thanks for watching. We will see you
next time on gun Jog Nature
Popular Podcasts
Hey Jonas!
Hey Jonas! The official Jonas Brothers podcast. Hosted by Kevin, Joe, and Nick Jonas. It’s the Jonas Brothers you know... musicians, actors, and well, yes, brothers. Now, they’re sharing another side of themselves in the playful, intimate, and irreverent way only they can. Spend time with the Jonas Brothers here and stay a little bit longer for deep conversations like never before.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by Audiochuck Media Company.
Betrayal Weekly
Betrayal Weekly is back for a new season. Every Thursday, Betrayal Weekly shares first-hand accounts of broken trust, shocking deceptions, and the trail of destruction they leave behind. Hosted by Andrea Gunning, this weekly ongoing series digs into real-life stories of betrayal and the aftermath. From stories of double lives to dark discoveries, these are cautionary tales and accounts of resilience against all odds. From the producers of the critically acclaimed Betrayal series, Betrayal Weekly drops new episodes every Thursday. If you would like to share your story, you can reach out to the Betrayal Team by emailing them at betrayalpod@gmail.com and follow us on Instagram at @betrayalpod and @glasspodcasts. Please join our Substack for additional exclusive content, curated book recommendations, and community discussions. Sign up FREE by clicking this link Beyond Betrayal Substack. Join our community dedicated to truth, resilience, and healing. Your voice matters! Be a part of our Betrayal journey on Substack.