157. Are You Making These Expensive AI Mistakes?

Send us a text! (Your number stays private)

AI is everywhere these days, and while it can be an amazing tool, it's also becoming a minefield of costly AI mistakes for business owners. In this episode, I’m pulling back the curtain on some of the most common missteps I’m seeing digital product sellers make when using AI. From letting AI tools write your legal policies (please don’t!) to blindly trusting AI-generated website code, these mistakes might seem small but can lead to big problems with your credibility, performance, or even legal compliance.

I walk you through why you should never use AI for things like privacy policies or contracts, and why AI code (though convenient) can easily break your site if you don’t fully understand what it’s doing. I even share a behind-the-scenes example of how my own excitement about an AI-built automation backfired. The bottom line? If you’re not double-checking AI’s output or editing it to reflect your actual brand voice, you could be hurting your business more than helping it.

This episode isn’t anti-AI - far from it! I use it all the time. But we need to be savvy about how we’re using it. So if you’ve been tempted to copy-paste AI copy, take its “facts” at face value, or let it take the reins on your tech without oversight, this is your nudge to pause and rethink your approach. Tune in to hear what to avoid today, and come back next week for the creative, strategic ways I do recommend using AI in your business.

01:24 - Using AI for legal content like privacy policies or contracts

04:00 - Letting AI generate website code without understanding it

06:52 - Trusting AI-generated “facts” without verifying

08:34 - Copy-pasting AI-generated copy without editing for brand voice

Links & Resources:

• Termageddon
• WordPress Care Plan
• Episode 148, The 15-Minute Weekly Plan: How AI Helped Me Stop Wasting My Best Brain Hours
• Follow me on Instagram @kristendoyle.co 
• Check out my Everything Page: a one-stop shop for savvy selling!
• The Savvy Seller Collective
• Join my private Facebook community: Savvy Teacher Sellers
• More resources for growing your TPT business
• Rate & review The Savvy Teacher Seller on Apple Podcasts

Show Notes: https://kristendoyle.co/episode157 

Feeling overwhelmed by all the things you think you need to do to grow your digital product business? Take the Savvy Seller Quiz and find out exactly what to focus on right now to actually move your business forward. 

 ➡️ Take the quiz at savvyquiz.com

Check out my Everything Page at https://kristendoyle.co/everything

Episode Transcript

Kristen (00:01):
AI is everywhere right now, and I know everyone is
talking about obvious uses, likewriting copy with AI or using it
to create images. But I amseeing business owners make some
pretty big and possiblyseriously damaging mistakes with

(00:23):
AI. So today, I am sharing someAI mistakes that could hurt your
website, they could tank yourcredibility, and they could even
land you in legal trouble. Sowhether you are just curious
about AI, and maybe you've beena little hesitant to start using
it, or you're already using itevery single day, this episode

(00:44):
could save your business fromsome costly problems down the
road.
Are you a digital product orcourse creator, selling on
platforms like Teachers PayTeachers, Etsy, or your own
website? Ready to grow yourbusiness, but not into the kind
of constant hustle that leadsstraight to burnout? Then you're

(01:04):
in the right place. Welcome toThe Savvy Seller. I'm Kristen
Doyle, and I'm here to give youno-fluff tools and strategies
that move the needle for yourbusiness without burning you out
in the process—things like SEO,no stress marketing, email list
building, automations and somuch more. Let's get started,

(01:24):
y'all.
So what shouldn't you do withAI? Here are some mistakes that
can actually cost you. The firstone is never, ever, ever, use AI
for legal stuff. Now, I knowit's tempting, you can ask chat

(01:46):
GPT for a privacy policy, andboom, there it is, ready to copy
paste onto your website. Butevery single word in legal
policies matters. That's thewhole point. Right? The words in
the policy or in the contract,are supposed to protect you when
things go wrong, or when someonethinks things went wrong, but

(02:07):
they really didn't. The goal isthat your contract or your
privacy policy or whatever it iscovers you.
See, the problem with AI is theymay not know what state you live
in or what country you're in.
They may not know what toolsyou're using on your website
when you draft that privacypolicy, or what different laws
apply to you for thatindependent contractor agreement

(02:29):
you had them draft.
Let's talk about privacypolicies for your website as an
example. Your privacy policyneeds different things depending
on what's on your website. Whatplugins are you using? Do you
have Google fonts loading onyour site? What email provider
do you have? Do you have aFacebook pixel or Google

(02:50):
Analytics tracking code or otherthings like that installed? All
sorts of little details likethat really matter in terms of
what has to go in your privacypolicy. So when you use an AI
tool, even if you try to tellthem all the things that you
have installed, chances areyou're not going to know what
all is important, because you'renot an attorney, or at least I'm

(03:12):
assuming you aren't. I'mcertainly not.
I use a tool called Termageddonfor my website policies, and I
offer it to my websitemaintenance plan clients as
well, because it actually keepsup with changing laws and
updates the policiesautomatically. It will send a
question out when a policy haschanged or when a law has

(03:32):
changed, and I can just answer acouple of questions in normal
English that I do understand,and then they will update my
legal policies on my website tomatch my answers for the
questions and to be compliantwith those new laws. So
definitely look into things likethat, as opposed to using chat
GPT or even just getting astatic policy that was written

(03:56):
by an attorney, because thenthose don't change, and these
laws are changing constantly.
Alright, the second one is onethat my web designer friends and
I are seeing a lot of lately.
See, people hear that chat GPTcan help with website code. And
yeah, it can. You can ask chatGPT, or any other AI tool, to
change a button color orwhatever, and it will spit out

(04:19):
code for you and tell you whereto put it. But the problem is,
if you don't understand thecode, then you are basically
playing Russian roulette withyour website. I've had clients
tell me so very proudly, I usechat GPT, and I fixed this
thing! But then they call mebecause now three other things

(04:41):
are broken, and they don't knowhow to fix it. So mistake number
two is, don't let AI code yourwebsite unless you really know
and understand the code thatit's giving you.
See what happens is, you getcode from chat GPT to fix one
problem, but it causes adifferent problem that you now

(05:02):
don't know how to solve. Or youkeep adding little chat GPT code
snippets all over your website,and your site starts running
like molasses because it'sbloated with junk code, and that
could tank your Google ratingbecause of performance issues.
I get it. I've been there too. Iactually just very recently had

(05:22):
a similar experience, not withwebsite code, because that's the
thing I'm an expert in. But Iwas so excited when I figured
out how to create thisautomation in Make that had been
giving me trouble. And I was soproud of myself. I had used AI
to help build out this wholecomplex thing, and it was
working, but I mentioned it toan automations expert friend of

(05:43):
mine, and he basically said, ohmy goodness, you built this in
the least efficient waypossible. And he helped me
rebuild it properly, so now itis running so much better.
The difference is, withautomations, if I mess that up,
I'm kind of the only oneaffected, at least with that
particular automation, becauseit wasn't automating sending

(06:04):
emails to people or anythinglike that, but with your website
code, you could break things forevery single visitor. You could
hurt your search rankings. Youcould create security
vulnerabilities, and it's justreally not a risk worth taking
just to save a few bucks.
Honestly, most of the time whenI have to go in and clean up
that chat GPT code, what happensis that it has caused so many

(06:28):
problems that it would have beencheaper and faster for the
client to have just asked me orasked whoever your web designer
is to do it right the firsttime. If you're on one of my
WordPress care plans, thoselittle fixes sometimes are
completely free, or they might,if they are charged, then they
would probably be super cheap,way better than having to ask me

(06:50):
to come in and fix somethinglater on.
Okay, this next one is big, big,big. I use chat GPT and Claude
all the time. I am not anti-AIat all. In fact, next week's
episode is going to be all aboutways to use it. But don't trust
AI for facts, like ever. I can'ttell you how many times AI

(07:12):
tools, chat GPT, Claude, othershave confidently, absolutely
sworn to me, even when Iquestioned them, that something
was true when it was completelyand totally false. They will
tell you a setting exists in anapp when it doesn't. They'll
give you step by stepinstructions to do something
that don't work. They will statefacts that are just completely

(07:36):
made up. And the kicker is theydon't even say, I think, or
maybe you could try, they say itwith complete confidence, like
they are absolutely certain,even when you question them.
I've had chat GPT tell me to,you know, go find a certain
setting in a specific locationin an app, or walk me through
exactly where to click. Go here,click this, look for this, click

(07:57):
that, and I go to the app, andthat setting just doesn't exist
anywhere. And when I go back andtell it, hey, that doesn't
exist, it apologizes and givesme more wrong information
sometimes.
So as a rule, anything that AItells you that's supposed to be
fact, you really need to goverify on your own. Don't just

(08:18):
take its word for it, no matterhow confident it sounds, even if
you have, in the past, said,Hey, please don't ever make
things up anymore. Just tell meyou don't know, or ask me more
questions or whatever. Because,trust me, I have done that so
many times, and it just doesn'tstick.
Alright, last but definitely notleast, do not, do not, do not

(08:39):
copy paste AI generated textcopy exactly word for word. If
you're like me, you can spot AIwriting from a mile away at this
point. You guys know, I'm sure,exactly what I'm talking about.
It's weirdly perfect, nevermakes a grammar mistake of any
kind, super polished kind oftone that sounds not like any

(09:03):
human ever talks. If you ask AIto write something and then you
just copy paste it word forword, it's gonna sound robotic
like that, because AI is arobot. Is it getting better?
Yes. Can you train it on yourvoice? Yes.
But even if you think thatyou've trained it well, and the
AI has totally nailed your brandvoice, it didn't, I promise. It

(09:28):
might be close, but a lot oftimes it misses little quirks,
personality touches, andespecially those little bits of
grammar mistakes that we allmake in our natural speech
patterns that we wouldn't makewhen we typed, and those are
some of the things that helppeople to connect with you as a
human, especially right now inthis world full of AI, let's

(09:49):
just be honest, AI junk all overthe place.
Maybe, you know, I veryfrequently use the phrase or the
word y'all I say, here's thething. A lot, and AI can
sometimes pick up on those, butit misses a lot of times the way
that I have of explaining thingsthat is uniquely mine. And it'll

(10:11):
miss that kind of stuff for you,too. And then there are also
always those dead giveaways thatscream, AI wrote me. One of them
lately is em dashes, that is thebigger dash. AI is obsessed with
em dashes. It uses themconstantly. And I have seen in
lots of places, people whoautomatically assume that

(10:33):
anything with an em dash in itis AI generated content.
Funny enough, I voice recordedsome brainstorm for this
episode, and when I put it intoAI, it put em dashes in this
very section where I'm talkingabout how it overuses em dashes.
And I will tell you next weekall about my process for how I

(10:54):
used AI to brainstorm thisepisode.
I actually love a good em dash.
I used to use them a lot, but Ihave started shifting to either
regular dashes or commas, orjust finding a different way to
phrase things, because AI hasmade them so overdone. And I'm
not the only one. Lots of otherpeople have said I feel like I
can't use an em dash anymorebecause AI uses them everywhere.

(11:15):
So always, always, always go inand mess with the copy a little
bit. Add your personality backin. You know, use that AI copy
as your starting point, but makeit sound like you actually wrote
it. My best advice for that isliterally read it out loud. And
if you feel weird saying it, youneed to edit some more. Because,

(11:37):
honestly, your audience can tellthe difference 100%. They don't
want to hear from some robotpretending to be you. They want
to hear from real, actual you.
I just talked about a coupleweeks ago, those trends that
we're seeing right now, peopleare wanting content from real
people that feels real, even ifthere's mistakes in it, because

(11:57):
they are tired of AI generated,I'm just going to call it what
it is again, AI generated junk.
Alright, here is what I want youto do this week. Think about how
you are currently using AI and Imean, really think about it and
be honest with yourself. Are youdoing any of these mistakes? Any
of these don't do things? Areyou copy pasting AI content

(12:21):
without editing it? Or maybe youonly edit one or two things and
you don't edit enough to make itreally sound like you? Are you
using chat GPT to add code toyour website when you really
don't understand what it'sdoing? Maybe you're trusting AI
facts without double checkingthem. If you are, zero judgment
from me, I promise. We have allbeen there, myself included.

(12:43):
But, and I say this with love,stop it. Before you try anything
new, just resolve not to keepdoing these things, making these
same mistakes. It will reallymake your business better and
stronger in the long run.
Alright, now that you know whatnot to do, in next week's
episode, I am gonna share kindof the flip side. I'm gonna

(13:04):
share some creative AI uses youprobably haven't tried,
including, like I just talkedabout, the exact way that I
planned these two episodes, thisone and the next one, using some
AI tools. You'll learn how touse AI without sounding like a
robot, and how to make sure yourcontent is unique to you and
your thoughts and perspectivesand ideas when everyone else's

(13:28):
content sounds exactly the samebecause they use the same AI
robot to write it.
Plus, I'll share some reallyhelpful ways to save yourself
time, and a couple of just kindof fun things that I've been
doing with AI in my personallife too. Make sure you click
the follow button so you don'tmiss next week's episode, and
while you're there, it wouldreally make my day if you left a
quick reading or a review forthe show. I'll talk to you soon.