July 8, 2025 • 44 mins
Join the Tool Use Discord: https://discord.gg/PnEGyXpjaX
David Cross reveals his step-by-step process for generating emotionally resonant AI content that avoids the uncanny valley. Discover the AI tools and strategies to build your own content factory, including using a custom GPT for scripts , ElevenLabs for realistic AI voice generation , and Midjourney for unique, hand-created style artwork. David breaks down his method for iterating on AI output, the importance of a human-in-the-loop to ensure quality , and how to use AI to create multilingual content for a global audience. This is your guide to leveling up your AI content and making something that truly connects with people.
Connect with David Cross:
Website: https://davidcross.com
LinkedIn: https://www.linkedin.com/in/ssorcdivad/
Kids Chatterbox Project: http://kidschatterbox.com/
Get FREE AI tools pip install tool-use-ai
Connect with us
https://x.com/ToolUseAI
https://x.com/MikeBirdTech
00:00:00 - Intro
00:02:38 - The AI Content Creation Workflow
00:15:22 - Cost And Time For An AI Podcast
00:18:38 - How To Take AI Content From Good To Great
00:28:18 - Making Your AI Content Multilingual
00:40:38 - Advice for Marketers in the AI Age
Subscribe for more insights on AI tools, productivity, and creativity.
Tool Use is a weekly conversation with AI experts brought to you by Anetic.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
It's not even uncanny valley. It's a sense of disbelief that
they're not really sure that they can trust their own senses.
And as you're going through the process of producing content,
you need to iteratively say, howdoes this feel?
How is the messaging? Does it resonate with people?
Or does it feel that we're just using AI for AI sake?
Welcome to episode 47 of tool Use, the weekly conversation
(00:23):
about AI tools and strategies brought to you by an Eddie.
I'm Mike Bird, and today we're talking about AI for content
creation. The tips and tricks to we'll let
anyone make high quality content.
This week we're joined by David Cross, a senior content marketer
and creator of the Kids Shatterbox podcast, Antika, the
100% personalized, on demand multilingual mindfulness app.
David, welcome to all use. Hey, Mike, thanks very much.
(00:44):
I'm glad to be here. Would.
You mind give us a bit of your background and how you got into
working with AI? So I got into AI about 3 1/2
years ago. I just started learning about
it. I was fascinating.
My background is that I've been in marketing for 35 years and in
the middle of the 90s, when the Internet first became popular,
(01:05):
I'd already been using bulletin boards and things with an old
modem. And I just discovered this thing
called the Internet in about 93 and I started learning about it.
I was there when the first Netscape browser started and by
sort of 94, I realised that the Internet was going to be such a
pivotal change for all our livesand it was going to bring in so
(01:28):
many different technologies and things into our lives.
And so I started to learn a lot about the Internet at that time.
And to me, AI now is analogous to what the Internet was back in
the early to mid 90s, that it's something that everyone should
be learning more about. It's a very cool and new
(01:49):
technology. And so there's still some kind
of warts and bumps with it, but I started learning about it and
learning about generative AI andhow I could use AI in terms of
my work as a marketer and how wecould start to bend and utilise
those technologies in in many different ways.
(02:11):
And I've used it every day for the last three years.
Excellent. Well, I'd love to dive into your
process for actually creating a podcast.
I use notebook LM to kind of take some documents, create a
little conversation, but it's very obviously an AI
conversation. It's not something that you can
really like squeak by. But but you know, we're getting
into one issue people have is the the quality of the output.
(02:33):
And when I checked out kids Chatterbox, it's it's
phenomenal. Like it sounds like a real story
and goes through with it. So you might tell us your
process of how you actually makean episode and how you
iteratively got to the high standard that is today.
Yeah, sure. I think 28 years in content
marketing helps. And I did voice over for many
years. I worked with companies like
American Express, Amazon, Nike, Hewlett Packard, etcetera, and I
(02:58):
did voice over for them. So my background in content
marketing and my understanding and background in voice over
gave me a kind of quality lens through which I would look.
And so kids Chatterbox started about two months ago.
My wife and I have a couple of smaller children and they love
(03:20):
stories, both stories that we read to them and also stories
like kids podcasts and kids videos on YouTube.
And we were just chatting and I said to my wife, I wonder if
it's possible to use AI, completely AI, to generate a
kids podcast or kids story channel.
And this is how my career has always gone, that I come up with
(03:45):
a kind of an inquisitive idea. I wonder if we could do this and
then I find a way to do it. And so the acid test for
creating a story that would be engaging for children would be
that it would have to be engaging and resonate
emotionally with them in the same way that if content had
(04:06):
been created by a human being, it would be a good story.
They'd be engaged. Parents would feel that it was
engaging and that it didn't havethat kind of uncanny valley feel
to you, which you mentioned notebook LM, the generative AI
notebook LM is pretty cool. You can feed it, you know, just
a basically a minimal PDF and it'll generate that two person
(04:30):
pseudo podcast example. And it's, it's pretty cool, but
it does sound like it's a littlebit generated.
And so I started experimenting with various AI text to speech
models commercially available and open source.
And I also created a pretty richprompt using first of all, a two
(04:57):
or three page branding document,which I created the same as I
would if I was working with a freelance writer or an agency.
And that described what these podcast stories would be about,
the sort of message in them, thecast of characters, the kind of
(05:17):
vision and framework of what those stories would be like.
Each story had to be cute and endearing for three to five year
olds, which was our target audience.
It would also have to be something that would include and
stories about childhood and stories about playing in the
dirt and, you know, making cupcakes out of pan cones and
(05:39):
things like that and getting dirty and would explore the joy
of just childhood play. And also that there would be a
little lesson in each one that the children could listen to the
to the stories. And it would also include a
little sort of a nod to A to a lesson in that.
And so I put together this branding document.
(06:01):
I also iteratively worked through different prompts to
create the stories and then whenI got it to about 90% of where I
wanted the prompt to be, then I put it together and created a
custom GPT that would ask me questions about OK, what's this
story about and who are the maincharacters and what theme do you
(06:24):
want to bring up and what stories are there.
Then it produces about usually 12 to 1500 words, which I then
download and edit in Word or Google Docs or something.
And as I'm editing that, I'm using my content creation
experience. And so I'm listening in my own
mind to that kind of internal voice reading these stories to
(06:47):
me. And like, is this a story that
my kids would find interesting? And then when I've got it just
as I wanted, then I take it to the next step, which is to use a
text to speech model. And the one I selected is it for
this project is 11 labs. And the reason is that 11 labs
to me was the one that was able to produce the audio, the the
(07:13):
voice that sounded most emotionally engaging and most
human like I suppose I would describe it as.
And then so we got the content, the voice over.
I then used Suno to generate a kind of kids podcast kind of
music and then edited in a 15 seconds of that at the start and
(07:36):
the end. And then I worked with Mid
Journey to develop a particular artistic style.
And if you look at the images and the YouTube videos on the
kids Chatterbox channel on YouTube, they looked hand
created. They don't look like something
that an AI generated. And so with every part of this
(08:00):
process, I was always thinking, does this feel like a human
created? Does this feel like it's
endearing and engaging for children?
The great thing is we have a couple of kids that would listen
to these test stories and you know, right from the beginning
they were engaged, they enjoyed these stories.
(08:20):
And so we had great feedback from the, from the very start
from, from our kids. The other thing which I found
really interesting was that 11 Labs has a feature within there
called the dubbing feature, which allows you to dub audio
from one language into another. And myself, I have, I've lived
(08:46):
in eight countries and in our family overall there are 5
nationalities represented, both with children, grandchildren and
partners of of my children. And I myself have worked in 21
countries and I've also worked with people on projects from
over 40 countries. And so it was really important
(09:09):
to me that we could bring these stories to a more of an
international audience. And so I contacted a couple of
my friends in India and friends in France and Sweden, and I did
a couple of tests where we translated one of these stories,
kids chatterbox stories, into those other languages.
(09:29):
And I sent it to them. And I said, does this capture
the right feeling? Does it feel like someone that's
reading an engaging story that kids would find, you know,
acceptable? And they, they all said, yeah,
it absolutely is. And so we then regionalized Kids
Chatterbox. So there's now English, Arabic,
(09:51):
Hindi, French, Spanish, Portuguese and Swedish.
And so Kids Chatterbox is now multilingual in seven languages.
And I registered the different channels on Spotify and on
YouTube. And So what started as a kind of
(10:13):
I wonder if grew into something really, I think it's pretty cool
that these stories are there andyou know when when we send them
out to people did well, they hadthe same reaction that you did.
There's no sense of uncanny valley.
They feel like like a human created these kids and adults
(10:33):
love the stories. People have said is that your
wife's voice or who did the artwork?
It's beautiful. And we don't get any sense from
anyone that there's a kind of robot in the background kind of
thing. We still love reading stories to
our kids and our kids very oftenon a car journey or at bedtime,
(10:53):
they'll say, could I have this story?
And so the stories, for example,where we live, it's a new
housing development, but there'salso some construction equipment
quite close by. And there's this old leaky water
truck that whenever we go a walk, my kids like to look at
this leaky water truck. And so we, the first, the first
(11:15):
episode is called Wally the Leaky Water truck.
And then there was a game I usedto play with my daughters when
they were younger called the pizza game.
And the pizza game we'd often played on a car trip or at
bedtime. And the pizza game was you can
select a crust and then red or white sauce.
(11:36):
And then you have three ingredients that you can put on.
And so we did this episode on the pizza game.
And then one of the other episodes is called Weather the
Weather. And I live in the Pacific
Northwest, and we have this saying here that there's never
any bad weather. There's only bad clothing.
And so we put this story together.
(11:57):
And then there was one story called A Hammer and a Helping
Hand. And when I was a kid, I would
think it was about 3 or 4. I remember my dad saying, you
know, that knots made wood weaker and so knots were not a
good thing. And I remember thinking about
what he'd said and then I noticed there was a knot in my
bedroom door. And so I went and found a hammer
(12:19):
and chisel from his, from his tool bagging chisel that this
knot in the middle of my bedroomdoor.
And so we we turned that into a kids chatterbox story.
And so we now have 5 stories in each of seven different
languages. So we, we've got 35 different
podcast stories, all holy AI generated.
(12:43):
And the 6th 1 is going to be outnext week.
And but it's just a really fun project.
And then as I was doing kids Chatterbox, people were saying,
you're really on to something here.
And I hadn't yet decided what that what that something was,
but it became kind of apparent what that was.
(13:06):
And that was the the next thing that you alluded to.
So I developed a prompt that would generate A5 to 10 minute
initially self hypnosis script based on whatever reason someone
would input into a form. Like I want to sleep better, I
(13:29):
want to feel calmer, I want to stop smoking or biting my nails
or whatever. And then it would generate a
script based on that and the specific instructions I put in
the prompt and then again it would feed it off, this time
via. I put together a scenario in
make.com that then sends that tothe 11 labs API endpoint.
(13:55):
So the form submits to a web hook, then it submits to the
open AI API, then that goes to the 11 labs and then it emails
the finished generated audio to someone.
So that was kind of interesting.But then I got to thinking,
well, it doesn't have to just beself hypnosis.
It could be a guided meditation or a sleep support or it could
(14:20):
be a change of habit is another one that we put together.
And so this now does 5 differentthings.
And it turns out that my my daughter-in-law is a software
engineer and we just got talkingand she is also a yoga teacher.
(14:41):
And she's like, this would make a really good app.
And so we're putting this together now we're working on on
designing the app, but we're also putting together a pitch
deck to try and find investors for this.
It all grew out of a discussion,a kind of what if discussion one
Saturday morning where my wife and I just said, I wonder if,
(15:04):
and it it just kind of grew, grew from that, from just
following the path of inquisitiveness and seeing where
this naturally, naturally took us.
And it was, I think it's pretty cool way, the way it's evolved.
Excellent. And it's very important to just
follow curiosity, play with things, explore.
One question I had is so a lot of people have the misconception
(15:26):
they kind of just put a single prompt in and get out a high
quality result. And that's why I let people get
disappointed with AI content creation because the baseline is
just very subpar. And with you being very human in
the loop, auditing the output, doing a little tweaks.
Can you give a general time frame for a 10 minute episode?
How much time you put into it and what the general cost is
using the tools like Mid Journey, Suno, Chatsu, BT?
I did a lot of playing around the 1st 10 days to try and
(15:50):
refine those ideas into a product or an audio podcast that
I think is acceptable. So if you put all the hours I
spent at night playing with it as sort of R&D and just cast
that aside. We got the 1st 2 episodes out
including all the generation on open AIS, API, 11 Labs audio,
(16:18):
mid journey subscription, the whole thing first two cost is
$82 and we've now put out 35 individual podcasts along with
we've currently done the first two episodes on video on
YouTube, including all the production, the whole thing for
that. So 35 podcasts plus 14 YouTube.
(16:42):
I think it's like $482. The entire cost that we we've
spent on it now obviously we don't need to bringing a
consultant to a software developer because I've all kind
of hacked it together myself. But that's that's the raw cost
that that we've spent on it pretty inexpensive.
And what what, what about the time commitment per episode are
(17:04):
you like, if, if you were to make the entire thing from
scratch, would this save you half the time or even more than
that? As I said, I worked in voice
over for a long time and I also worked, I've been a podcast host
myself and I know how long thosethings can take to actually
produce it. So in terms of generating the
story it takes, it takes about 5minutes for me to draft out the
(17:30):
specific story ideas and the things I want to include.
Then it takes you know, 30 seconds for ChatGPT to generate
that. Then I edit it.
So I would probably say half an hour for the actual, the
content, the story. Then inside of 11 labs, I would
say that it took a couple of minutes generation and then I go
(17:53):
through to it and listen and just tweak it here and there.
In terms of regenerating individual paragraphs mid
journey, a few moments. I would say that it's probably
if I added it up between 90 minutes and two hours per
episode in total pretty quick. Yeah, absolutely.
And one thing I wanted to touch on is a lot of people almost
(18:14):
dismiss children's content as being less complex, easier to
generate. But when we talked previously,
you brought up how kids will just focus on whatever grabs
their attention the most. So they just get honed in on
something. It has to be really well
composed with kids. Chatterbox.
It's not like the TikTok style dopamine hacking.
Just try to like capture with hyperstimulation.
It's just like nice and and melodic and it's it's a really
(18:36):
enjoyable experience. Outside of just the experience
you have and the taste that you've acquired over that time
for generating something. Do you have any tips or tricks
for people to take their contentfrom good to great?
Are there certain heuristics they can follow?
Are there certain patterns they can look out for?
Or how can I take something thatAI spits out at me and then go
through a process of tweaking itso it becomes a higher standard
of quality? Yeah, sure.
(18:57):
So I mean you mentioned there that that it's not the same
content, for example, as TikTok.It's not.
You'll notice in the videos, forexample, that we use still
images with a Ken Burns effect rather than video.
And people have asked me like, why did you not use AI video to
generate that? Well, for our target audience
(19:20):
who are small children, there are so many stimulating videos
around at the moment and so manystimulating things.
And what we wanted for this particular podcast was to have
something that when parents need10 minutes to themselves, they
could get their kids to listen to this and it would settle them
down and it wouldn't further stimulate them.
(19:41):
So we consciously chose for our audience to not use very flashy
or very stimulating videos. I think the thing with content
creation is you have to put yourself in the, in the mind or
the ears or the eyes of your listener, your audience and see
(20:03):
whether it fits within that framework.
So if your audience is in our case someone that you didn't
want to over stimulate, you wouldn't put that in your video.
And I think it's like from the vast array of all the tools that
we have. I mean, VO3 just came out and
that's incredible, some of the stuff you can do with that.
(20:24):
I think that the content you produce has to be congruent
first of all, with the audience that you're trying to pitch that
content to. And although you can do some
things like we could have done asoundtrack that was like my
three-year old son loves the Ramones.
We could have used the Ramones, for example, as the music or
something very, very driven and fast, but it wouldn't be
(20:47):
suitable for our target audience.
So I think that you need to select, first of all, select the
tools and the content that is most suitable to your audience.
And then I think you need to take a step back and also
include a few beta testers who are naturally more sceptic and
less wooed by technology and that are more able to to say,
(21:11):
yeah, that that feels good. And that was one of our acid
tests. When we sent it to people, the
early tests, we said, how does this feel?
Not what do you think about it, but how does it feel?
And the feedback was like, this feels great.
And I think that the same applies with a lot of content is
(21:33):
that it has to have the right, the right feeling, the right
connection. You know, I think that with the
preponderance of AI tools, I remember years ago when digital
cameras first started and there was this feeling that we
shouldn't be using digital. What was wrong with analogue?
Ansel Adams never needed digital, etcetera.
(21:54):
And so, you know, digital photography now is commonplace.
But then there was the move someyears ago to like vinyl became
more popular and cassette tapes and and analogue cameras.
And there was that move back towards that.
And you find that the more AI that's available and the more
(22:16):
cool things and technology that's available, there is a
need for people to feel greater connection to things that they
can trust with their own senses.Because as they start to
question, is that real? That there was an advert years
ago for Memorex and there was this the slogan, is it live or
(22:38):
is it Memorex? Because Memorex faithfully
reproduced audio to the extent that you wouldn't know whether
it was live or Memorex. And so the same thing is very
true of, of AI that it's like, is that real or is it AI?
And people have got that. It's not even uncanny valley.
It's a sense of disbelief that they're not really sure that
they can trust their own senses.And so I would say that you got
(23:02):
to be, you got to be congruent to your audience and, and to
deliver content that feels realistic for your audience.
And as you're going through the process of producing content, I
think that you need to iteratively say, how does this
feel? How is the messaging?
(23:22):
How is the content itself? Does it resonate with people or
does it feel that we're just using technology for
technology's sake or using AI for AI sake?
I think it needs to be even morereal and connected with people
the more technology you use. So I would base a lot of it on
(23:43):
how does this feel in terms of your, in terms of your target
audience? I like that a lot.
I think it's it's extremely important to bring the human
aspect into it and not a lot of the technology just overwhelm
the entire process. Do you have any advice for
people who are trying to build up that sense of taste where
they may consume a certain degree of content?
(24:05):
Some they like, some they dislike, but when they start
making their own, it just doesn't have that right feeling
and they don't know where to take it.
Yeah. So I would bring in some, I
mean, for example, customers or a user base that's
representative of your customersthat matches the demographics.
For example, I have a friend in the UK who has an essential oils
(24:28):
company and his customer base isabout 85% women between the ages
of about 27 and 58. And so, you know, he'll very
often ask me to give ideas and recommendations.
And I'll, my first recommendation is to ask Mary,
(24:49):
his wife to come into the conversation because he is an
Oxford trained. Actually, he has a degree in in
psychology and he is able to produce things that are quite
well delineated, but don't always capture the right
feeling. And then it's very important
(25:11):
that you bring in and involve your customers.
And so if you're marketing to, for example, a user base of of
that's predominantly female and yet everyone in the room is a
man and that you have some competent programmers, but you
don't have that feeling elements.
And so I would say bring in yourcustomer base and, and involve
(25:32):
them and ask them for feedback. And does this answer your
question? And in your company, for
example, you'll very often find that the customer service and
that the people that answer the telephones have the finger on
the pulse of what customers actually want and what they need
to answer the questions. And so if you're just a single
(25:54):
user sat behind your computer trying to generate this content,
bringing more people that are users and consumers of your
content, like for me, it's very easy.
I just went through to the kitchen and said, hey, kids,
what do you think of this? And if they were engaged, then
it was easy. But involving those who are the
(26:15):
recipients and the target audience of your content is
always a good thing to do. And I'd recommend that you
branch out into into bringing people in to, if not necessarily
collaborate on your content. At least if you say, hey, can I
just show you a few minutes of this?
And you're not interested in whether they think it's cool.
(26:35):
You're interested in whether it resonates with them, whether it
feels right, whether it's, you know, if it, if you're doing
like for example, an AI driven podcast on science, does it come
across as a clear and cogent explanation of microbial DNA or
(26:56):
something like that? Does it like if you go back to
some of those great documentaries that the BBC did
many years ago, for example, there's one of my favorite on,
on Richard Feynman called the pleasure of finding things out.
And if you use that as an example of great content,
Feynman is a Nobel Prize winning, award-winning
(27:20):
physicist, like super, super intellectually sharp, but he was
able to teach and convey that knowledge in a very engaging,
endearing, humorous and, and, and brilliant way.
He was a brilliant teacher. And I think that, you know, all
content has to have that human connection, regardless of what
(27:44):
it is. And by involving people in that
content, you're going to find out whether it actually
resonates with them, no matter what it's about.
I think that's great insight. A few people on the show have
come on and talked about the importance of talking to the
user, but breaking it down to how they're actually feeling and
how it's resonating, I think it's extremely important.
One thing you've done with Kids Chatterbox that I absolutely
(28:06):
love is, is making a multilingual allow it to reach
more people for this. It just allows more children to
experience the magic. But from a business perspective,
it makes a lot of sense to just increase your total addressable
market. Try to hit as many people as
possible. So what is the process of once
you hit that good enough marketing or piece of content to
then bring it to an international audience?
(28:26):
How do you do you just use like change the toggle on 11 labs or
is there something deeper to it?Yeah.
So the way that 11 Labs works isthat you take your, in my case,
English or when I say English, Ialways think of American and
British English because it was Ithink George Bernard Shaw that
(28:47):
said that we were two countries separated by a common language.
And so you take your English version and then you upload it
into the 11 Lads Dubbings dubbing area in the user
interface and you select your target language that you want to
work with. And then you wait a few moments
(29:09):
and what it does first of all isit goes through and it
translates the text. And so you can see a side by
side translation and then it generates the audio.
And I mean, for example, I do speak Swedish, but I don't
really speak Spanish and I certainly don't speak Hindi or
(29:29):
Arabic. And so the question comes, well,
how did you know it sounded right?
So first of all, I did some tests and sent it off to people
that I know that speak Hindi andFrench and Spanish and said, how
does this sound? How does it feel?
And my goal was not perfect university quality Spanish.
(29:51):
I didn't want it to be perfect. What I wanted it to be was like,
you know, if you go to a different country where they
don't speak English, but you still have a go at the language
and there's still the ability tocommunicate with people.
I wanted it to be at least that good, but more importantly, to
feel engaging, tend to still generate the same feeling.
(30:15):
And so I, I sent out audio teststo friends and I said, you know,
is the language good enough? But does it feel right?
Does it still convey that feeling?
And so when I generate audios now, I'll still go through and
listen to the Hindi, the Arabic,the French versions and I'm
looking for, first of all, are there any hallucinations, for
(30:40):
example, when you generate the audio?
I mean, 11 Labs is great, but it's still an AI system.
And so you can get some unusual sounding things, for example, in
the kids Chatterbox audio, we'd get nice narration and that
suddenly it would get very high and you could immediately hear
it. And so I would just go back and
regenerate that paragraph. But in, for example, the Arabic
(31:02):
version, Arabic is a, a beautiful language.
And when I listen to the Arabic version, first of all, I'm
listening to those kind of weird, high pitched, you know,
off the wall generations. And I'll regenerate that, but
I'm closing my eyes and I'm listening to, I'm like, how does
this feel? There's this feel, right?
(31:23):
And so I use that same test to assess Arabic and Hindi as I
would for English, not the wordsthemselves, but how does the
feeling feel? And I think that's actually a
pretty good, a pretty good test.And my inability to speak or
(31:47):
understand Arabic or Hindi, it seems didn't matter because
using that tool faithfully translated the words enough.
But the fact that I was using the same test of how does it
feel when I sent it to some friends in India, for example,
did speak Hindi, they were able to say, yeah, it feels right.
(32:08):
And so that's an interesting test.
Even though I didn't speak that language, I could still use
that, that measure of how does it feel?
And so we would generate the audio.
And then I use the same template.
I use Pro Tools, but you could use Garage Band or any audio
(32:30):
editor and I'd simply insert theArabic or Hindi or French or
Swedish text, sorry, audio in place of the English one.
And then just make sure that thein music comes in and fades in,
fades out and then fades back inat the end and then publish it
(32:50):
on. I use Spotify as the podcast
platform and then and then publish it.
And we have, as I say, all 5 episodes in seven different
languages. And it's been fun.
It's been really fun. And it's been really insightful
for learning about how to make an entire piece of content with
AI. I'm curious if you have any
words of wisdom or advice for the average content creator who
(33:12):
already has a workflow of of making their own videos, maybe
make their own music, but they want to supplement it with AI.
Do you have any recommendations for tools, workflows, or just
general advice for how they can enhance existing content with
AII? Think the first thing is to, to,
to become inquisitive again, because sometimes you get into
creating content and you know, you're particular media or
(33:35):
medium. So you can maybe publishing
content on Instagram, for example, short videos and you
get a little bit entrenched and you need to kind of think
outside the box. And So what I would encourage
people to do is start playing with tools just as as I had done
and seeing what you can do with them.
(33:57):
And then from that you'll be able to then apply those to your
existing workflows. You can also take for example, I
was talking to a guy yesterday who's a video producer for a
large financial publisher. And what he wanted to do was to
clone some of their editors voices to create custom voices,
(34:18):
which he can do in in 11 labs ina number of a number of other
systems. And then start to use those
cloned voices to generate audio where they would have normally
gone to the expert themselves and says, can you record this?
They can now enter text and havethe cloned AI voice read that
(34:39):
content. So that's already kind of built
into or being built into their workflow.
I would say also like ask yourself inquisitive questions.
I wonder if AI could do this. I wonder if we could substitute
this. I worked with someone who's a
content creator and she does photography on Instagram.
(35:04):
And so I helped her generate a what's called the Laura model
for Lora model that she now usesto generate some images of
herself that she uses in Instagram posts.
And you cannot tell the difference between you know,
(35:27):
what's what's real and what she's actually generated.
And like just there are some excellent courses on LinkedIn.
There's a guy called Dave Burst BIRSS who has some great courses
on LinkedIn about a generative AI.
Just go out there and Start learning.
(35:47):
I mean, as I say, I remember the, the Internet when it first
became popular in the mid 90s. I just turned 60 last November.
And like, I don't feel it. I feel that this is such a, a
new and vibrant time and that everyone should be learning
about AI and what AI can do and the possibilities.
(36:08):
And like just learn about it andget inquisitive and start
thinking how could I use it? Is the one thing I would say is
really important regardless of what you do, whether you're an
accountant, an attorney, a content creator, video producer,
whatever it is software developer.
For example, I've been testing various tools to help me with my
(36:31):
fledgling code writing as part of the Tika app project that
I've been working on. And so just get inquisitive, I
think is is the answer to that question.
Yeah, I, I fully agree between curiosity and just taking some
agency for exploring the the world's our oyster at this day
and age. I'm curious, with your
(36:53):
exploration and you know, inquisitive nature, where do you
think the current gaps are in AIfor content creation that either
new tools will solve or forever will be the human touch?
Yeah, I think that it's to do not so much with the tool, but
the way it's being used. So, you know, the, the thing I
(37:15):
was like to tell people when they're using any tool is that
you've got to stay clear of the tool until you know what you
want to do with it. And that applies to, for
example, whether you're going tocook a meal.
You need to first of all have the meal in your consciousness
before you start chopping vegetables.
(37:36):
You it's the same with any content creation.
It's the same with software development.
I'm sure you mean you're a software developer.
I would not pass myself as more than a basic hacker, but it's
the thinking behind the use of atool.
And there's that old adage of measure twice, cut once, and so
you have to apply that thinking.So I would say that the speed at
(38:00):
which AI tools will continue to evolve will get faster and
faster, and we'll be able to do more with it.
But the thinking behind the AI tool is what needs to evolve.
There was a study that came out last week which showed that
(38:21):
people that relied on AI became a bit kind of, I suppose the way
of, of paraphrasing is that theybecame more lazy and complacent
in terms of, oh, I'll just let the AI do that.
But that's not how I like to seethe AI tools.
I like to think through the process, think through who my
target audience is, be consciously aware of what I want
(38:45):
the tool to do, then start usingthe tool.
So I always say before you startmessing around with any tool,
stay away from the tool and decide what it is you want to do
with it. And then the same is applicable
to AI now as it was to the Internet in the mid 1990s.
You've got to to think through the process and that goes to
(39:09):
businesses. For example, I remember when SEO
became really quite important for businesses and you know, a
lot of businesses realise that it was not that difficult to
generate traffic from SEO, but hey, now the person's here, what
are we going to do with them? So you've got to think of not
just getting them there, but in terms of the conversion
(39:32):
mechanisms and the whole life cycle of the customers of how
you're going to move them on from that initial inquiry to
convert them to maybe a free report or a trial use.
And how you then going to monetise that in terms of your
overall workflow. And that has nothing to do with
tools. That has to do with considering
(39:53):
what you want the outcome to be from that.
So the answer from my side is not that the AI tools are
lacking. They will, they will grow and
they will evolve and develop. But I think it's the.
The lack of critical thinking that is going to be a potential
downfall for many individuals and businesses here because they
(40:17):
don't consider what they they actually want to do with these
tools beforehand. And last question for me, I feel
like you've touched on in a little bit a lot of your
answers, but I'd love just more laying it all out.
You have a lot of experience marketing and today's age there
is more noise than ever before. So being able to get a message
across is getting harder than ever, yet we're enabled with
these tools. Do you have any advice for
(40:38):
marketers to thrive in the AIH? Yeah, sure.
You know, I come from a, what wecall a direct response marketing
background. I'm not so much of A brand
marketer. There was I think believe he was
AUS Postmaster General many years ago called John Wanamaker.
And he famously said that I knowthat half of my advertising
(41:01):
dollars are wasted. I just don't know which half.
And The thing is with a direct response marketing mindset is
that you're always thinking of the ability to quantify what
what it is you've done from a marketing perspective.
So for example, if you see a billboard when you're driving
(41:22):
down the road, a few rare cases if you use a specific phone
number and stuff, but you don't really know how many people have
converted from that. Whereas for example, in direct
response marketing, in direct mail, if you send a specific
letter to someone in the mail, you know the effect that that
particular letter has. And so when the Internet first
came about, the thing that we realised pretty quickly was that
(41:46):
human beings had not changed very much.
Although the technology was new,the humans were, you know, we,
we evolved over much slower timeframes.
This is why a geneticist use fruit flies because the the life
cycle is only a day long and so they can see genetic changes
quickly. But humans take many millennia
(42:07):
to evolve. But the the thing to realise is
that in the AI, in digital age, the human beings, our customers
have not really changed from whothey were as humans 100 years
ago. And so the ability to connect
with people and the ability to connect with your customers
(42:28):
requires even a greater authenticity in, in your
communication, authenticity in the ability to create that
connection. Connection will become more
important the more that we look at our screens and social media
to get some form of validation of who we are and aren't I
(42:49):
popular, I got, you know, 10,000likes and stuff.
The fact is that people are really becoming more
disconnected the more technologythey use and consume and the
ability to be more congruent andmore connected and authentic
with our customers is even more important.
So I would be, without going into specifics because it
(43:11):
depends on the individual business, but I would be looking
for ways to become more deeply connected while using
technologies like AI in a way that creates a more authentic
and real connection with prospects and with customers.
David, this was a wonderful conversation.
Thank you for sharing your time and afterward tease with us.
Before we let you go, is there anything you'd like the audience
(43:32):
to know? I, I think I would, I not so
much know, but I, I think that go from this podcast and, and
start experimenting, start becoming inquisitive, start
rolling up your sleeves and trying new things and, and you
know, AI can seem a little bit scary at first.
Take one of the introductory courses on generative AI on
(43:55):
LinkedIn, for example, from thisguy Dave Burse, who I think is
really good at what he does, andjust Start learning more about
AI and then start becoming inquisitive of how could I apply
this to my profession? How could I apply it to my
personal life? How could I use these
technologies? If you've any questions after
(44:15):
listening to this, please feel free to reach out.
I'd love to continue the discussion with people, but get
inquisitive and and start following that inquisitiveness
and see where it leads because something cool may happen.
You may be invited on podcast. You never know.
All right, dude, have a great day.
It's not even uncanny valley. It's a sense of disbelief that
they're not really sure that they can trust their own senses.
And as you're going through the process of producing content,
you need to iteratively say, howdoes this feel?
How is the messaging? Does it resonate with people?
Or does it feel that we're just using AI for AI sake?
Welcome to episode 47 of tool Use, the weekly conversation
(00:23):
about AI tools and strategies brought to you by an Eddie.
I'm Mike Bird, and today we're talking about AI for content
creation. The tips and tricks to we'll let
anyone make high quality content.
This week we're joined by David Cross, a senior content marketer
and creator of the Kids Shatterbox podcast, Antika, the
100% personalized, on demand multilingual mindfulness app.
David, welcome to all use. Hey, Mike, thanks very much.
(00:44):
I'm glad to be here. Would.
You mind give us a bit of your background and how you got into
working with AI? So I got into AI about 3 1/2
years ago. I just started learning about
it. I was fascinating.
My background is that I've been in marketing for 35 years and in
the middle of the 90s, when the Internet first became popular,
(01:05):
I'd already been using bulletin boards and things with an old
modem. And I just discovered this thing
called the Internet in about 93 and I started learning about it.
I was there when the first Netscape browser started and by
sort of 94, I realised that the Internet was going to be such a
pivotal change for all our livesand it was going to bring in so
(01:28):
many different technologies and things into our lives.
And so I started to learn a lot about the Internet at that time.
And to me, AI now is analogous to what the Internet was back in
the early to mid 90s, that it's something that everyone should
be learning more about. It's a very cool and new
(01:49):
technology. And so there's still some kind
of warts and bumps with it, but I started learning about it and
learning about generative AI andhow I could use AI in terms of
my work as a marketer and how wecould start to bend and utilise
those technologies in in many different ways.
(02:11):
And I've used it every day for the last three years.
Excellent. Well, I'd love to dive into your
process for actually creating a podcast.
I use notebook LM to kind of take some documents, create a
little conversation, but it's very obviously an AI
conversation. It's not something that you can
really like squeak by. But but you know, we're getting
into one issue people have is the the quality of the output.
(02:33):
And when I checked out kids Chatterbox, it's it's
phenomenal. Like it sounds like a real story
and goes through with it. So you might tell us your
process of how you actually makean episode and how you
iteratively got to the high standard that is today.
Yeah, sure. I think 28 years in content
marketing helps. And I did voice over for many
years. I worked with companies like
American Express, Amazon, Nike, Hewlett Packard, etcetera, and I
(02:58):
did voice over for them. So my background in content
marketing and my understanding and background in voice over
gave me a kind of quality lens through which I would look.
And so kids Chatterbox started about two months ago.
My wife and I have a couple of smaller children and they love
(03:20):
stories, both stories that we read to them and also stories
like kids podcasts and kids videos on YouTube.
And we were just chatting and I said to my wife, I wonder if
it's possible to use AI, completely AI, to generate a
kids podcast or kids story channel.
And this is how my career has always gone, that I come up with
(03:45):
a kind of an inquisitive idea. I wonder if we could do this and
then I find a way to do it. And so the acid test for
creating a story that would be engaging for children would be
that it would have to be engaging and resonate
emotionally with them in the same way that if content had
(04:06):
been created by a human being, it would be a good story.
They'd be engaged. Parents would feel that it was
engaging and that it didn't havethat kind of uncanny valley feel
to you, which you mentioned notebook LM, the generative AI
notebook LM is pretty cool. You can feed it, you know, just
a basically a minimal PDF and it'll generate that two person
(04:30):
pseudo podcast example. And it's, it's pretty cool, but
it does sound like it's a littlebit generated.
And so I started experimenting with various AI text to speech
models commercially available and open source.
And I also created a pretty richprompt using first of all, a two
(04:57):
or three page branding document,which I created the same as I
would if I was working with a freelance writer or an agency.
And that described what these podcast stories would be about,
the sort of message in them, thecast of characters, the kind of
(05:17):
vision and framework of what those stories would be like.
Each story had to be cute and endearing for three to five year
olds, which was our target audience.
It would also have to be something that would include and
stories about childhood and stories about playing in the
dirt and, you know, making cupcakes out of pan cones and
(05:39):
things like that and getting dirty and would explore the joy
of just childhood play. And also that there would be a
little lesson in each one that the children could listen to the
to the stories. And it would also include a
little sort of a nod to A to a lesson in that.
And so I put together this branding document.
(06:01):
I also iteratively worked through different prompts to
create the stories and then whenI got it to about 90% of where I
wanted the prompt to be, then I put it together and created a
custom GPT that would ask me questions about OK, what's this
story about and who are the maincharacters and what theme do you
(06:24):
want to bring up and what stories are there.
Then it produces about usually 12 to 1500 words, which I then
download and edit in Word or Google Docs or something.
And as I'm editing that, I'm using my content creation
experience. And so I'm listening in my own
mind to that kind of internal voice reading these stories to
(06:47):
me. And like, is this a story that
my kids would find interesting? And then when I've got it just
as I wanted, then I take it to the next step, which is to use a
text to speech model. And the one I selected is it for
this project is 11 labs. And the reason is that 11 labs
to me was the one that was able to produce the audio, the the
(07:13):
voice that sounded most emotionally engaging and most
human like I suppose I would describe it as.
And then so we got the content, the voice over.
I then used Suno to generate a kind of kids podcast kind of
music and then edited in a 15 seconds of that at the start and
(07:36):
the end. And then I worked with Mid
Journey to develop a particular artistic style.
And if you look at the images and the YouTube videos on the
kids Chatterbox channel on YouTube, they looked hand
created. They don't look like something
that an AI generated. And so with every part of this
(08:00):
process, I was always thinking, does this feel like a human
created? Does this feel like it's
endearing and engaging for children?
The great thing is we have a couple of kids that would listen
to these test stories and you know, right from the beginning
they were engaged, they enjoyed these stories.
(08:20):
And so we had great feedback from the, from the very start
from, from our kids. The other thing which I found
really interesting was that 11 Labs has a feature within there
called the dubbing feature, which allows you to dub audio
from one language into another. And myself, I have, I've lived
(08:46):
in eight countries and in our family overall there are 5
nationalities represented, both with children, grandchildren and
partners of of my children. And I myself have worked in 21
countries and I've also worked with people on projects from
over 40 countries. And so it was really important
(09:09):
to me that we could bring these stories to a more of an
international audience. And so I contacted a couple of
my friends in India and friends in France and Sweden, and I did
a couple of tests where we translated one of these stories,
kids chatterbox stories, into those other languages.
(09:29):
And I sent it to them. And I said, does this capture
the right feeling? Does it feel like someone that's
reading an engaging story that kids would find, you know,
acceptable? And they, they all said, yeah,
it absolutely is. And so we then regionalized Kids
Chatterbox. So there's now English, Arabic,
(09:51):
Hindi, French, Spanish, Portuguese and Swedish.
And so Kids Chatterbox is now multilingual in seven languages.
And I registered the different channels on Spotify and on
YouTube. And So what started as a kind of
(10:13):
I wonder if grew into something really, I think it's pretty cool
that these stories are there andyou know when when we send them
out to people did well, they hadthe same reaction that you did.
There's no sense of uncanny valley.
They feel like like a human created these kids and adults
(10:33):
love the stories. People have said is that your
wife's voice or who did the artwork?
It's beautiful. And we don't get any sense from
anyone that there's a kind of robot in the background kind of
thing. We still love reading stories to
our kids and our kids very oftenon a car journey or at bedtime,
(10:53):
they'll say, could I have this story?
And so the stories, for example,where we live, it's a new
housing development, but there'salso some construction equipment
quite close by. And there's this old leaky water
truck that whenever we go a walk, my kids like to look at
this leaky water truck. And so we, the first, the first
(11:15):
episode is called Wally the Leaky Water truck.
And then there was a game I usedto play with my daughters when
they were younger called the pizza game.
And the pizza game we'd often played on a car trip or at
bedtime. And the pizza game was you can
select a crust and then red or white sauce.
(11:36):
And then you have three ingredients that you can put on.
And so we did this episode on the pizza game.
And then one of the other episodes is called Weather the
Weather. And I live in the Pacific
Northwest, and we have this saying here that there's never
any bad weather. There's only bad clothing.
And so we put this story together.
(11:57):
And then there was one story called A Hammer and a Helping
Hand. And when I was a kid, I would
think it was about 3 or 4. I remember my dad saying, you
know, that knots made wood weaker and so knots were not a
good thing. And I remember thinking about
what he'd said and then I noticed there was a knot in my
bedroom door. And so I went and found a hammer
(12:19):
and chisel from his, from his tool bagging chisel that this
knot in the middle of my bedroomdoor.
And so we we turned that into a kids chatterbox story.
And so we now have 5 stories in each of seven different
languages. So we, we've got 35 different
podcast stories, all holy AI generated.
(12:43):
And the 6th 1 is going to be outnext week.
And but it's just a really fun project.
And then as I was doing kids Chatterbox, people were saying,
you're really on to something here.
And I hadn't yet decided what that what that something was,
but it became kind of apparent what that was.
(13:06):
And that was the the next thing that you alluded to.
So I developed a prompt that would generate A5 to 10 minute
initially self hypnosis script based on whatever reason someone
would input into a form. Like I want to sleep better, I
(13:29):
want to feel calmer, I want to stop smoking or biting my nails
or whatever. And then it would generate a
script based on that and the specific instructions I put in
the prompt and then again it would feed it off, this time
via. I put together a scenario in
make.com that then sends that tothe 11 labs API endpoint.
(13:55):
So the form submits to a web hook, then it submits to the
open AI API, then that goes to the 11 labs and then it emails
the finished generated audio to someone.
So that was kind of interesting.But then I got to thinking,
well, it doesn't have to just beself hypnosis.
It could be a guided meditation or a sleep support or it could
(14:20):
be a change of habit is another one that we put together.
And so this now does 5 differentthings.
And it turns out that my my daughter-in-law is a software
engineer and we just got talkingand she is also a yoga teacher.
(14:41):
And she's like, this would make a really good app.
And so we're putting this together now we're working on on
designing the app, but we're also putting together a pitch
deck to try and find investors for this.
It all grew out of a discussion,a kind of what if discussion one
Saturday morning where my wife and I just said, I wonder if,
(15:04):
and it it just kind of grew, grew from that, from just
following the path of inquisitiveness and seeing where
this naturally, naturally took us.
And it was, I think it's pretty cool way, the way it's evolved.
Excellent. And it's very important to just
follow curiosity, play with things, explore.
One question I had is so a lot of people have the misconception
(15:26):
they kind of just put a single prompt in and get out a high
quality result. And that's why I let people get
disappointed with AI content creation because the baseline is
just very subpar. And with you being very human in
the loop, auditing the output, doing a little tweaks.
Can you give a general time frame for a 10 minute episode?
How much time you put into it and what the general cost is
using the tools like Mid Journey, Suno, Chatsu, BT?
I did a lot of playing around the 1st 10 days to try and
(15:50):
refine those ideas into a product or an audio podcast that
I think is acceptable. So if you put all the hours I
spent at night playing with it as sort of R&D and just cast
that aside. We got the 1st 2 episodes out
including all the generation on open AIS, API, 11 Labs audio,
(16:18):
mid journey subscription, the whole thing first two cost is
$82 and we've now put out 35 individual podcasts along with
we've currently done the first two episodes on video on
YouTube, including all the production, the whole thing for
that. So 35 podcasts plus 14 YouTube.
(16:42):
I think it's like $482. The entire cost that we we've
spent on it now obviously we don't need to bringing a
consultant to a software developer because I've all kind
of hacked it together myself. But that's that's the raw cost
that that we've spent on it pretty inexpensive.
And what what, what about the time commitment per episode are
(17:04):
you like, if, if you were to make the entire thing from
scratch, would this save you half the time or even more than
that? As I said, I worked in voice
over for a long time and I also worked, I've been a podcast host
myself and I know how long thosethings can take to actually
produce it. So in terms of generating the
story it takes, it takes about 5minutes for me to draft out the
(17:30):
specific story ideas and the things I want to include.
Then it takes you know, 30 seconds for ChatGPT to generate
that. Then I edit it.
So I would probably say half an hour for the actual, the
content, the story. Then inside of 11 labs, I would
say that it took a couple of minutes generation and then I go
(17:53):
through to it and listen and just tweak it here and there.
In terms of regenerating individual paragraphs mid
journey, a few moments. I would say that it's probably
if I added it up between 90 minutes and two hours per
episode in total pretty quick. Yeah, absolutely.
And one thing I wanted to touch on is a lot of people almost
(18:14):
dismiss children's content as being less complex, easier to
generate. But when we talked previously,
you brought up how kids will just focus on whatever grabs
their attention the most. So they just get honed in on
something. It has to be really well
composed with kids. Chatterbox.
It's not like the TikTok style dopamine hacking.
Just try to like capture with hyperstimulation.
It's just like nice and and melodic and it's it's a really
(18:36):
enjoyable experience. Outside of just the experience
you have and the taste that you've acquired over that time
for generating something. Do you have any tips or tricks
for people to take their contentfrom good to great?
Are there certain heuristics they can follow?
Are there certain patterns they can look out for?
Or how can I take something thatAI spits out at me and then go
through a process of tweaking itso it becomes a higher standard
of quality? Yeah, sure.
(18:57):
So I mean you mentioned there that that it's not the same
content, for example, as TikTok.It's not.
You'll notice in the videos, forexample, that we use still
images with a Ken Burns effect rather than video.
And people have asked me like, why did you not use AI video to
generate that? Well, for our target audience
(19:20):
who are small children, there are so many stimulating videos
around at the moment and so manystimulating things.
And what we wanted for this particular podcast was to have
something that when parents need10 minutes to themselves, they
could get their kids to listen to this and it would settle them
down and it wouldn't further stimulate them.
(19:41):
So we consciously chose for our audience to not use very flashy
or very stimulating videos. I think the thing with content
creation is you have to put yourself in the, in the mind or
the ears or the eyes of your listener, your audience and see
(20:03):
whether it fits within that framework.
So if your audience is in our case someone that you didn't
want to over stimulate, you wouldn't put that in your video.
And I think it's like from the vast array of all the tools that
we have. I mean, VO3 just came out and
that's incredible, some of the stuff you can do with that.
(20:24):
I think that the content you produce has to be congruent
first of all, with the audience that you're trying to pitch that
content to. And although you can do some
things like we could have done asoundtrack that was like my
three-year old son loves the Ramones.
We could have used the Ramones, for example, as the music or
something very, very driven and fast, but it wouldn't be
(20:47):
suitable for our target audience.
So I think that you need to select, first of all, select the
tools and the content that is most suitable to your audience.
And then I think you need to take a step back and also
include a few beta testers who are naturally more sceptic and
less wooed by technology and that are more able to to say,
(21:11):
yeah, that that feels good. And that was one of our acid
tests. When we sent it to people, the
early tests, we said, how does this feel?
Not what do you think about it, but how does it feel?
And the feedback was like, this feels great.
And I think that the same applies with a lot of content is
(21:33):
that it has to have the right, the right feeling, the right
connection. You know, I think that with the
preponderance of AI tools, I remember years ago when digital
cameras first started and there was this feeling that we
shouldn't be using digital. What was wrong with analogue?
Ansel Adams never needed digital, etcetera.
(21:54):
And so, you know, digital photography now is commonplace.
But then there was the move someyears ago to like vinyl became
more popular and cassette tapes and and analogue cameras.
And there was that move back towards that.
And you find that the more AI that's available and the more
(22:16):
cool things and technology that's available, there is a
need for people to feel greater connection to things that they
can trust with their own senses.Because as they start to
question, is that real? That there was an advert years
ago for Memorex and there was this the slogan, is it live or
(22:38):
is it Memorex? Because Memorex faithfully
reproduced audio to the extent that you wouldn't know whether
it was live or Memorex. And so the same thing is very
true of, of AI that it's like, is that real or is it AI?
And people have got that. It's not even uncanny valley.
It's a sense of disbelief that they're not really sure that
they can trust their own senses.And so I would say that you got
(23:02):
to be, you got to be congruent to your audience and, and to
deliver content that feels realistic for your audience.
And as you're going through the process of producing content, I
think that you need to iteratively say, how does this
feel? How is the messaging?
(23:22):
How is the content itself? Does it resonate with people or
does it feel that we're just using technology for
technology's sake or using AI for AI sake?
I think it needs to be even morereal and connected with people
the more technology you use. So I would base a lot of it on
(23:43):
how does this feel in terms of your, in terms of your target
audience? I like that a lot.
I think it's it's extremely important to bring the human
aspect into it and not a lot of the technology just overwhelm
the entire process. Do you have any advice for
people who are trying to build up that sense of taste where
they may consume a certain degree of content?
(24:05):
Some they like, some they dislike, but when they start
making their own, it just doesn't have that right feeling
and they don't know where to take it.
Yeah. So I would bring in some, I
mean, for example, customers or a user base that's
representative of your customersthat matches the demographics.
For example, I have a friend in the UK who has an essential oils
(24:28):
company and his customer base isabout 85% women between the ages
of about 27 and 58. And so, you know, he'll very
often ask me to give ideas and recommendations.
And I'll, my first recommendation is to ask Mary,
(24:49):
his wife to come into the conversation because he is an
Oxford trained. Actually, he has a degree in in
psychology and he is able to produce things that are quite
well delineated, but don't always capture the right
feeling. And then it's very important
(25:11):
that you bring in and involve your customers.
And so if you're marketing to, for example, a user base of of
that's predominantly female and yet everyone in the room is a
man and that you have some competent programmers, but you
don't have that feeling elements.
And so I would say bring in yourcustomer base and, and involve
(25:32):
them and ask them for feedback. And does this answer your
question? And in your company, for
example, you'll very often find that the customer service and
that the people that answer the telephones have the finger on
the pulse of what customers actually want and what they need
to answer the questions. And so if you're just a single
(25:54):
user sat behind your computer trying to generate this content,
bringing more people that are users and consumers of your
content, like for me, it's very easy.
I just went through to the kitchen and said, hey, kids,
what do you think of this? And if they were engaged, then
it was easy. But involving those who are the
(26:15):
recipients and the target audience of your content is
always a good thing to do. And I'd recommend that you
branch out into into bringing people in to, if not necessarily
collaborate on your content. At least if you say, hey, can I
just show you a few minutes of this?
And you're not interested in whether they think it's cool.
(26:35):
You're interested in whether it resonates with them, whether it
feels right, whether it's, you know, if it, if you're doing
like for example, an AI driven podcast on science, does it come
across as a clear and cogent explanation of microbial DNA or
(26:56):
something like that? Does it like if you go back to
some of those great documentaries that the BBC did
many years ago, for example, there's one of my favorite on,
on Richard Feynman called the pleasure of finding things out.
And if you use that as an example of great content,
Feynman is a Nobel Prize winning, award-winning
(27:20):
physicist, like super, super intellectually sharp, but he was
able to teach and convey that knowledge in a very engaging,
endearing, humorous and, and, and brilliant way.
He was a brilliant teacher. And I think that, you know, all
content has to have that human connection, regardless of what
(27:44):
it is. And by involving people in that
content, you're going to find out whether it actually
resonates with them, no matter what it's about.
I think that's great insight. A few people on the show have
come on and talked about the importance of talking to the
user, but breaking it down to how they're actually feeling and
how it's resonating, I think it's extremely important.
One thing you've done with Kids Chatterbox that I absolutely
(28:06):
love is, is making a multilingual allow it to reach
more people for this. It just allows more children to
experience the magic. But from a business perspective,
it makes a lot of sense to just increase your total addressable
market. Try to hit as many people as
possible. So what is the process of once
you hit that good enough marketing or piece of content to
then bring it to an international audience?
(28:26):
How do you do you just use like change the toggle on 11 labs or
is there something deeper to it?Yeah.
So the way that 11 Labs works isthat you take your, in my case,
English or when I say English, Ialways think of American and
British English because it was Ithink George Bernard Shaw that
(28:47):
said that we were two countries separated by a common language.
And so you take your English version and then you upload it
into the 11 Lads Dubbings dubbing area in the user
interface and you select your target language that you want to
work with. And then you wait a few moments
(29:09):
and what it does first of all isit goes through and it
translates the text. And so you can see a side by
side translation and then it generates the audio.
And I mean, for example, I do speak Swedish, but I don't
really speak Spanish and I certainly don't speak Hindi or
(29:29):
Arabic. And so the question comes, well,
how did you know it sounded right?
So first of all, I did some tests and sent it off to people
that I know that speak Hindi andFrench and Spanish and said, how
does this sound? How does it feel?
And my goal was not perfect university quality Spanish.
(29:51):
I didn't want it to be perfect. What I wanted it to be was like,
you know, if you go to a different country where they
don't speak English, but you still have a go at the language
and there's still the ability tocommunicate with people.
I wanted it to be at least that good, but more importantly, to
feel engaging, tend to still generate the same feeling.
(30:15):
And so I, I sent out audio teststo friends and I said, you know,
is the language good enough? But does it feel right?
Does it still convey that feeling?
And so when I generate audios now, I'll still go through and
listen to the Hindi, the Arabic,the French versions and I'm
looking for, first of all, are there any hallucinations, for
(30:40):
example, when you generate the audio?
I mean, 11 Labs is great, but it's still an AI system.
And so you can get some unusual sounding things, for example, in
the kids Chatterbox audio, we'd get nice narration and that
suddenly it would get very high and you could immediately hear
it. And so I would just go back and
regenerate that paragraph. But in, for example, the Arabic
(31:02):
version, Arabic is a, a beautiful language.
And when I listen to the Arabic version, first of all, I'm
listening to those kind of weird, high pitched, you know,
off the wall generations. And I'll regenerate that, but
I'm closing my eyes and I'm listening to, I'm like, how does
this feel? There's this feel, right?
(31:23):
And so I use that same test to assess Arabic and Hindi as I
would for English, not the wordsthemselves, but how does the
feeling feel? And I think that's actually a
pretty good, a pretty good test.And my inability to speak or
(31:47):
understand Arabic or Hindi, it seems didn't matter because
using that tool faithfully translated the words enough.
But the fact that I was using the same test of how does it
feel when I sent it to some friends in India, for example,
did speak Hindi, they were able to say, yeah, it feels right.
(32:08):
And so that's an interesting test.
Even though I didn't speak that language, I could still use
that, that measure of how does it feel?
And so we would generate the audio.
And then I use the same template.
I use Pro Tools, but you could use Garage Band or any audio
(32:30):
editor and I'd simply insert theArabic or Hindi or French or
Swedish text, sorry, audio in place of the English one.
And then just make sure that thein music comes in and fades in,
fades out and then fades back inat the end and then publish it
(32:50):
on. I use Spotify as the podcast
platform and then and then publish it.
And we have, as I say, all 5 episodes in seven different
languages. And it's been fun.
It's been really fun. And it's been really insightful
for learning about how to make an entire piece of content with
AI. I'm curious if you have any
words of wisdom or advice for the average content creator who
(33:12):
already has a workflow of of making their own videos, maybe
make their own music, but they want to supplement it with AI.
Do you have any recommendations for tools, workflows, or just
general advice for how they can enhance existing content with
AII? Think the first thing is to, to,
to become inquisitive again, because sometimes you get into
creating content and you know, you're particular media or
(33:35):
medium. So you can maybe publishing
content on Instagram, for example, short videos and you
get a little bit entrenched and you need to kind of think
outside the box. And So what I would encourage
people to do is start playing with tools just as as I had done
and seeing what you can do with them.
(33:57):
And then from that you'll be able to then apply those to your
existing workflows. You can also take for example, I
was talking to a guy yesterday who's a video producer for a
large financial publisher. And what he wanted to do was to
clone some of their editors voices to create custom voices,
(34:18):
which he can do in in 11 labs ina number of a number of other
systems. And then start to use those
cloned voices to generate audio where they would have normally
gone to the expert themselves and says, can you record this?
They can now enter text and havethe cloned AI voice read that
(34:39):
content. So that's already kind of built
into or being built into their workflow.
I would say also like ask yourself inquisitive questions.
I wonder if AI could do this. I wonder if we could substitute
this. I worked with someone who's a
content creator and she does photography on Instagram.
(35:04):
And so I helped her generate a what's called the Laura model
for Lora model that she now usesto generate some images of
herself that she uses in Instagram posts.
And you cannot tell the difference between you know,
(35:27):
what's what's real and what she's actually generated.
And like just there are some excellent courses on LinkedIn.
There's a guy called Dave Burst BIRSS who has some great courses
on LinkedIn about a generative AI.
Just go out there and Start learning.
(35:47):
I mean, as I say, I remember the, the Internet when it first
became popular in the mid 90s. I just turned 60 last November.
And like, I don't feel it. I feel that this is such a, a
new and vibrant time and that everyone should be learning
about AI and what AI can do and the possibilities.
(36:08):
And like just learn about it andget inquisitive and start
thinking how could I use it? Is the one thing I would say is
really important regardless of what you do, whether you're an
accountant, an attorney, a content creator, video producer,
whatever it is software developer.
For example, I've been testing various tools to help me with my
(36:31):
fledgling code writing as part of the Tika app project that
I've been working on. And so just get inquisitive, I
think is is the answer to that question.
Yeah, I, I fully agree between curiosity and just taking some
agency for exploring the the world's our oyster at this day
and age. I'm curious, with your
(36:53):
exploration and you know, inquisitive nature, where do you
think the current gaps are in AIfor content creation that either
new tools will solve or forever will be the human touch?
Yeah, I think that it's to do not so much with the tool, but
the way it's being used. So, you know, the, the thing I
(37:15):
was like to tell people when they're using any tool is that
you've got to stay clear of the tool until you know what you
want to do with it. And that applies to, for
example, whether you're going tocook a meal.
You need to first of all have the meal in your consciousness
before you start chopping vegetables.
(37:36):
You it's the same with any content creation.
It's the same with software development.
I'm sure you mean you're a software developer.
I would not pass myself as more than a basic hacker, but it's
the thinking behind the use of atool.
And there's that old adage of measure twice, cut once, and so
you have to apply that thinking.So I would say that the speed at
(38:00):
which AI tools will continue to evolve will get faster and
faster, and we'll be able to do more with it.
But the thinking behind the AI tool is what needs to evolve.
There was a study that came out last week which showed that
(38:21):
people that relied on AI became a bit kind of, I suppose the way
of, of paraphrasing is that theybecame more lazy and complacent
in terms of, oh, I'll just let the AI do that.
But that's not how I like to seethe AI tools.
I like to think through the process, think through who my
target audience is, be consciously aware of what I want
(38:45):
the tool to do, then start usingthe tool.
So I always say before you startmessing around with any tool,
stay away from the tool and decide what it is you want to do
with it. And then the same is applicable
to AI now as it was to the Internet in the mid 1990s.
You've got to to think through the process and that goes to
(39:09):
businesses. For example, I remember when SEO
became really quite important for businesses and you know, a
lot of businesses realise that it was not that difficult to
generate traffic from SEO, but hey, now the person's here, what
are we going to do with them? So you've got to think of not
just getting them there, but in terms of the conversion
(39:32):
mechanisms and the whole life cycle of the customers of how
you're going to move them on from that initial inquiry to
convert them to maybe a free report or a trial use.
And how you then going to monetise that in terms of your
overall workflow. And that has nothing to do with
tools. That has to do with considering
(39:53):
what you want the outcome to be from that.
So the answer from my side is not that the AI tools are
lacking. They will, they will grow and
they will evolve and develop. But I think it's the.
The lack of critical thinking that is going to be a potential
downfall for many individuals and businesses here because they
(40:17):
don't consider what they they actually want to do with these
tools beforehand. And last question for me, I feel
like you've touched on in a little bit a lot of your
answers, but I'd love just more laying it all out.
You have a lot of experience marketing and today's age there
is more noise than ever before. So being able to get a message
across is getting harder than ever, yet we're enabled with
these tools. Do you have any advice for
(40:38):
marketers to thrive in the AIH? Yeah, sure.
You know, I come from a, what wecall a direct response marketing
background. I'm not so much of A brand
marketer. There was I think believe he was
AUS Postmaster General many years ago called John Wanamaker.
And he famously said that I knowthat half of my advertising
(41:01):
dollars are wasted. I just don't know which half.
And The thing is with a direct response marketing mindset is
that you're always thinking of the ability to quantify what
what it is you've done from a marketing perspective.
So for example, if you see a billboard when you're driving
(41:22):
down the road, a few rare cases if you use a specific phone
number and stuff, but you don't really know how many people have
converted from that. Whereas for example, in direct
response marketing, in direct mail, if you send a specific
letter to someone in the mail, you know the effect that that
particular letter has. And so when the Internet first
came about, the thing that we realised pretty quickly was that
(41:46):
human beings had not changed very much.
Although the technology was new,the humans were, you know, we,
we evolved over much slower timeframes.
This is why a geneticist use fruit flies because the the life
cycle is only a day long and so they can see genetic changes
quickly. But humans take many millennia
(42:07):
to evolve. But the the thing to realise is
that in the AI, in digital age, the human beings, our customers
have not really changed from whothey were as humans 100 years
ago. And so the ability to connect
with people and the ability to connect with your customers
(42:28):
requires even a greater authenticity in, in your
communication, authenticity in the ability to create that
connection. Connection will become more
important the more that we look at our screens and social media
to get some form of validation of who we are and aren't I
(42:49):
popular, I got, you know, 10,000likes and stuff.
The fact is that people are really becoming more
disconnected the more technologythey use and consume and the
ability to be more congruent andmore connected and authentic
with our customers is even more important.
So I would be, without going into specifics because it
(43:11):
depends on the individual business, but I would be looking
for ways to become more deeply connected while using
technologies like AI in a way that creates a more authentic
and real connection with prospects and with customers.
David, this was a wonderful conversation.
Thank you for sharing your time and afterward tease with us.
Before we let you go, is there anything you'd like the audience
(43:32):
to know? I, I think I would, I not so
much know, but I, I think that go from this podcast and, and
start experimenting, start becoming inquisitive, start
rolling up your sleeves and trying new things and, and you
know, AI can seem a little bit scary at first.
Take one of the introductory courses on generative AI on
(43:55):
LinkedIn, for example, from thisguy Dave Burse, who I think is
really good at what he does, andjust Start learning more about
AI and then start becoming inquisitive of how could I apply
this to my profession? How could I apply it to my
personal life? How could I use these
technologies? If you've any questions after
(44:15):
listening to this, please feel free to reach out.
I'd love to continue the discussion with people, but get
inquisitive and and start following that inquisitiveness
and see where it leads because something cool may happen.
You may be invited on podcast. You never know.
All right, dude, have a great day.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.