Don't tell Mom you love her with ChatGPT – unless you are a robot
AI is cool, but it’s mostly hype – for now.
Yes, you can write all kinds of exciting things with AI. That’s certainly true. The next Faulkner probably won’t be a “prompt engineer” who figures out the best way to wrangle content out of ChatGPT (yes, that’s a thing), but he might well get his big idea from there. Or use it to draft an outline.
In the business world, I’m seeing all kinds of crazy stuff emerge about how AI is going to lead to less work for everyone or replace pastors in churches. The hype is out of control. Mark Precision Outreach down as “skeptical.”
This article will shed a bit of light on why the AI hype, especially in the context of marketing, is relatively unfounded. We’re going to review a fun example of how you could (but shouldn’t) write a letter to your mother with ChatGPT. Then, we’ll briefly touch on how to be smart with GPT to help your workflow in any business (see! it’s not all bad).
Google is not stupid
Much of what we deal with as a marketing company is the ridiculousness put out by other marketing companies. A lot of that has to do with “quick fixes,” “easy schemes,” or “tricking algorithms.” Sigh.
A lot of business owners I’ve spoken with are sick of marketers for this exact reason – there’s so much promised that it’s almost impossible to separate the actual deliverable from the rhetoric. Or is it the same? For many such companies, you bet it is.
Whether or not it deserves it, ChatGPT has wiggled its way squarely in the “marketing hype” category, along with other AI tools.
Here’s an example you’ll see a lot of. Many people seem to think that AI replaces the need for human beings to write content altogether. Isn’t that a thought? Instead of hiring US-based content creators like us, or even outsourcing this work to folks overseas, you can use a robot to write all your content! If only it were that easy.
If I’m honest, ChatGPT can produce some impressive content. Anyone who plays with it a bit comes away surprised – a computer can do that? Nevertheless, Google is way ahead of it. Here’s Google’s search guru John Mueller on Twitter about this exact question last year:
See what he says? “Low-quality content.”
So what do they count as low-quality content?
Google looks for topical authority when ranking pages. That means: does Google see your page as providing a level of helpful information and expertise on the subject matter being queried?
Granted, Google’s algorithms aren’t perfect. That doesn’t mean it’s wise to try and “trick the system” by loading a gazillion backlinks or doing what’s called “keyword stuffing.” Shady marketers do these sorts of tactics and fail.
There are legitimate ways to do the things both of those techniques try to do, and then there are spammy ways that Google hates. Lazy marketers tend to use the ways Google hates (and hates more ever year). Over the years, Google (which has access to more of the world’s information than anyone else, including about you) has figured out how to separate these out.
20 years ago, people wrote downright silly “articles” to try to rank. Even today, you’ll see examples where some “marketing expert” will write about plumbers in Kentucky with something like this: “We’re Plumbers in Kentucky! If you have toilet issues you need plumbers in Kentucky! Why are we plumbers in Kentucky? Because we are plumbers in Kentucky!” It’s more common than you think.
And it doesn’t work most of the time, not these days.
Google looks for good content. In fact, Google figures out pretty quickly if you have a variety of content and not just random articles written by someone in Indonesia (no offense to Indonesia). Pictures? Small plus. Videos? BIG plus. Really big plus. Google is getting better at figuring out when people put high-quality video content on their site, and helping that content.
Obviously, if Google shared how it did all of this, it wouldn’t be a 1.5 Trillion dollar company. Therefore, we have to keep in mind what we’re dealing with. Google says it wants high quality content – it’s a good idea to believe them. They’re not stupid.
Using AI alone to generate content is asking for trouble
Putting aside the fact that AI-generated content can expose you to significant legal liability (especially intellectual property and plagiarism concerns), from a content perspective, AI is just not what it’s often billed as – at least if you’re using only AI to generate content (as many lazy marketers are).
I’m going to show you a somewhat unrealistic, but quite illustrative, example (I bet people do this, though).
Let’s assume you asked ChatGPT to write a letter to your mother. I did just that.
I’ll address the value of the content in a moment, but let’s take a look at what a well-respected content checker called GPTZero – which professors use to see if their students are cheating by using AI – says about the letter.
Chat GPT letter to mom
My Dearest [Mother’s Name],
As I sit down to pen this letter, I am filled with an overwhelming sense of gratitude and love for the incredible woman you are. There are so many words that could describe you – caring, compassionate, strong, nurturing – but none of them truly capture the depth of what you mean to me. You are so much more than just a mother; you are my guiding light and the rock upon which I stand.
Throughout my life, you have been a constant source of love and support, providing a safe and nurturing haven in which I could grow and flourish. Your unwavering belief in me has been the driving force behind my endeavors, always pushing me to reach for the stars. Your encouragement has given me the confidence to face life’s challenges head-on, knowing that with you by my side, I am never alone.
Your boundless love and selflessness have taught me the true meaning of compassion and empathy. You have been a shoulder to cry on during my toughest times and a cheerleader celebrating my every success. Your presence in my life has made the journey all the more beautiful and meaningful.
I cherish the moments we’ve shared together, from the laughter-filled family gatherings to the quiet, intimate conversations we’ve had. Each memory is etched in my heart, forming a tapestry of love and warmth that I will carry with me forever.
You have not only been an exceptional mother but also an incredible role model. Your grace, integrity, and perseverance have shown me the kind of person I aspire to become. Your unconditional love has given me the strength to love myself and others unconditionally.
On this special day, I want to take the opportunity to say thank you. Thank you for your sacrifices, your understanding, and your constant love. You have given me the greatest gift in the world – the gift of your heart and soul.
No words can truly express how much you mean to me, but I hope this letter comes close. I love you beyond measure, and I am eternally grateful for the blessing of having you as my mother.
With all my love,
[Your Name]
GPTZero not impressed
GPTZero, however, was not thrilled.
It says that the sentences highlighted in yellow are suspect for AI use. Well…
As mentioned before, Google will almost certainly penalize content like this because it’s AI-generated.
We cannot know Google’s exact algorithms, but we know that Google is not going to let people spin up tons and tons of AI content without putting in any real work and then be declared an authority.
The problem isn’t in the amount of content or the actual data in the content – it’s how the content is written, lacks certain language cues, and overall just misses a human touch.
The most flattering fake letter to mom in the world
You’ve got to keep in mind what ChatGPT does if you are trying to use it to generate content. It does not create content from some sort of artificial brain – oh no. What it actually does is generate content from a synthesis of text information available on the internet prior to 2021.
Here’s what ZDNet reported about it: “ChatGPT’s power is the ability to parse queries and produce fully fleshed-out answers and results based on most of the world’s digitally-accessible text-based information – at least information that existed as of its time of training prior to 2021.”
It is still incredibly impressive. But it’s not entirely original, either.
No matter what you’ve heard, ChatGPT lacks the ability to think in any way similar to a human being. It’s not about to go and take over the world – it can’t even write a convincing brief letter!
Let’s understand why.
Paragraph one: “As I sit down to pen this letter, I am filled with an overwhelming sense of gratitude and love for the incredible woman you are. There are so many words that could describe you – caring, compassionate, strong, nurturing – but none of them truly capture the depth of what you mean to me. You are so much more than just a mother; you are my guiding light and the rock upon which I stand.”
As I sit down to pen this letter? Who says that!? Wait – I know who. If you read a lot of old books, or maybe historical fiction, you might as well. Ding ding ding! It’s one of those tropes used in old-school letters, and modern iterations therein. If you were playing Dungeons and Dragons, and had to write a letter, you might well say that. “As I sit down to pen this letter!”
In real life, no one talks like that. But ChatGPT does not understand that. It probably never will, because things like idioms, cultural contexts, and social dynamics are incredibly intuitive. When it takes the “brain trust” of written content the world over, in English, and tries to write a letter to mom, this sort of silly phrasing somehow rises to the top. GPT likely cannot know that no one actually speaks that way outside of jailed bards in England from 500 years ago.
I remember reading once that no matter how well a person learns a foreign language, they’ll never sound like a native if they can’t learn the common turns of phrase. If you speak Spanish with Mexicans (which we do a lot in Texas), you will find that many answer the phone with “Bueno!” Yet every language textbook you’ll find translates “bueno” as good, which it is. For Mexican Spanish speakers, it’s just another way to say hello. Will ChatGPT ever get stuff like that? How could it?
The other thing I want to shine a light on in that first paragraph is the detached coldness of it. GPT is merely throwing this stuff together from existing content and spinning it up into something unique, but it comes up with a good list of traits – mom is caring, compassionate, strong, nurturing. Look at how that list is introduced, though: “there are so many words that could describe you.”
I’ve seen that, again, in old-style letters, maybe cheesy romance letters, but it’s not how anyone actually writes. It also focuses on the fact that these traits are just words – and to GPT, they really are just words, pieces of information, the meaning of which is derived from an underlying definition and millions of phrases it can search in its digital database that use them.
If you saw an actual letter like this, you’d feel that something was off, right? It just seems… too much. I think people now say “extra.” So, it’s too extra.
Cold as A-ice
The entire letter feels detached and cold. There is no personality, no specific examples – how could there be? ChatGPT knows nothing about mother.
Now, presumably – cue those prompt engineers – you could feed it a bunch of specific pieces of information to derive a better letter. That’s what some big GPT proponents would tell you. So instead of write a letter to mom, it would be more like, “Write a letter to my mother, mentioning the time she waited for 10 hours in the snow on Black Friday to buy me laptop I used to ask you to write her this letter, ChatGPT”
Presumably that would work better – sort of. I’ve spent many hours working with GPT at this point, and one thing that strikes me is that it has little ability to draw relevant conclusions based on personal data. If you put in a ton of personal information hoping to get a customized result, GPT is not programmed to spin up all sorts of likely scenarios that might proceed from the data unless you ask it to. It may well do so, based upon the broad content, but it won’t know, again, the context unless you tell it.
Therein lies the problem. You can get some highly customized content out of GPT if you work hard enough at it, but you have to give it so much information to get there that you might as well write the thing yourself. Or hire someone – a real person – to write the content for you.
If you spend all the time you could spend on writing instead of coming up with ChatGPT prompts for marketing, it would have been better spent.
It’s not just a loss of time: although using AI might seem like a way to save money, it’s a lot cheaper to get the job right the first time. Imagine having your entire site bounced or ranked down because of AI content. That’s not worth the gamble, in my opinion.
How I learned to stop worrying and love the bomb AI
Robots are not going to take over the earth. I don’t think.
Here’s the metrics for dentist searches in Kingwood…
Contact
Schedule
Use this contact form to send us a message. We’ll get back to you ASAP. Or, you may schedule 30 minutes of our time with the scheduling app to the right.
You can schedule 15 or 30 minutes at no cost to talk with us about your needs below. We’re available 6 days per week at a time convenient for you. This is, by far, the most convenient (and helpful) way for us to see if we can help you out. By prior agreement or for existing clients, you may also book 1-hour meetings and phone consultations.
Contact
Use this contact form to send us a message. We’ll get back to you ASAP. Or, you may schedule 30 minutes of our time with the scheduling app to the right.
Schedule
You can schedule 15 or 30 minutes at no cost to talk with us about your needs below. We’re available 6 days per week at a time convenient for you. This is, by far, the most convenient (and helpful) way for us to see if we can help you out. By prior agreement or for existing clients, you may also book 1-hour meetings and phone consultations.