I use AI to write emails, create press release, generate graphics, create videos, create posters, flyers, create voice overs, and so many things that I need for marketing. I frequently use different kinds of AI tools to create marketing materials. What about you?

AI can be a game-changer for marketing productivity. I’ve heard a lot about using AI to automate marketing but I haven't tried it yet. Thanks for reminding me of this!

AI is revolutionizing the creation of marketing materials by streamlining processes and boosting creativity. With AI-powered tools, businesses can generate tailored content, from social media posts to ad copy, in a fraction of the time it used to take. AI’s ability to analyze data and predict trends helps marketers craft personalized campaigns that resonate with target audiences. Moreover, it enhances design automation, allowing for rapid production of visually appealing graphics and layouts.

AI is an incredible productivity tool, but the results you see from it will depend on the human elements you add to it. You cannot pick the content or results from an AI and send it out. You have to add the human element if you want the content to work for you.

 

However, AI can save you enough time that you have a few more minutes to polish or improve a marketing piece before putting it out.

The problem with AI is that it's usually very obvious that's what has been used to do the writing or the designing. For writing, there are words that show up often like "elevate", for instance. I think AI needs to expand its dictionary usage?

 

As for AI graphics, there is just something about them that doesn't feel right. It doesn't feel genuine in a way? I guess I am at a loss to explain it, but maybe it's that real artists can detect it and perhaps someday the look of the graphics will be less easy to spot.

The problem with AI is that it's usually very obvious that's what has been used to do the writing or the designing. For writing, there are words that show up often like "elevate", for instance. I think AI needs to expand its dictionary usage?

 

As for AI graphics, there is just something about them that doesn't feel right. It doesn't feel genuine in a way? I guess I am at a loss to explain it, but maybe it's that real artists can detect it and perhaps someday the look of the graphics will be less easy to spot.

 

I definitely agree with you on the point about AI graphics being super easy to spot. I play an online game that uses a lot of AI generated graphics in it's updates and press releases and it's absolutely so off-putting and ugly as hell in my opinion. Anyone even reasonably experienced with graphics & AI can likely tell the difference very quickly is my suspicion.

I think AI needs to expand its dictionary usage?

Sort of.

 

The problem is that words like elevate are less used normally and when they are used, they're statistically 'more interesting'. AI in the GPT sense doesn't actually 'know' anything, it doesn't really have a 'dictionary' in as much as it knows that certain words statistically go together enough to form a meaningful correlation - and just regurgitates from there.

 

Of course that's also not the whole story - the likes of Grammarly (which is also an AI product in the same vein) works off statistical relevance but in the opposite direction, when it talks about 'simplifying writing', it usually means trending towards what is statistically more common. The writing equivalent of writing elevator music vs a stunning concerto.

 

Then we have AI art tools and that's where it gets reeeeeaaaaaaally complicated. Yes, if you use Midjourney it's generated and there's a fair bet that it will be visibly noticeable. Depending on what was generated this may be more or less relevant to your situation. But, if you use Photoshop with Context-Aware Fill... you're leaning on many of the same techniques under the hood, and Photoshop has any number of tools in newest versions that are really AI-led without advertising that they are such.

 

I actually have a feeling that there's more marketing material out there that has AI at its heart than we currently realise because plenty of people will have started with ChatGPT and then sufficiently polished off the oddities to look close enough. That's the thing about the business world; people who can do something cheaper even if it's getting a computer to do 50% of the legwork first have an annoying habit of winning the race because the race is increasingly to the bottom.

Sort of.

 

The problem is that words like elevate are less used normally and when they are used, they're statistically 'more interesting'. AI in the GPT sense doesn't actually 'know' anything, it doesn't really have a 'dictionary' in as much as it knows that certain words statistically go together enough to form a meaningful correlation - and just regurgitates from there.

 

Of course that's also not the whole story - the likes of Grammarly (which is also an AI product in the same vein) works off statistical relevance but in the opposite direction, when it talks about 'simplifying writing', it usually means trending towards what is statistically more common. The writing equivalent of writing elevator music vs a stunning concerto.

 

Then we have AI art tools and that's where it gets reeeeeaaaaaaally complicated. Yes, if you use Midjourney it's generated and there's a fair bet that it will be visibly noticeable. Depending on what was generated this may be more or less relevant to your situation. But, if you use Photoshop with Context-Aware Fill... you're leaning on many of the same techniques under the hood, and Photoshop has any number of tools in newest versions that are really AI-led without advertising that they are such.

 

I actually have a feeling that there's more marketing material out there that has AI at its heart than we currently realise because plenty of people will have started with ChatGPT and then sufficiently polished off the oddities to look close enough. That's the thing about the business world; people who can do something cheaper even if it's getting a computer to do 50% of the legwork first have an annoying habit of winning the race because the race is increasingly to the bottom.

 

You really think that elevate is less used normally? Maybe we see different sites on the internet, but for the past 2 years it's been overkill to the point that I'll leave site after seeing it! It's become a huge pet peeve for me to see "elevate" because I know that people...and companies...like to hop on the word bandwagon. One year it was "populate" at the place where I worked.

 

What I've done if I use AI to write, is to go through and flag the words that are overused by other marketing/advertising/etc... and I'll get my thesaurus out and choose synonyms that might appeal to more audiences, those audiences who might also be sick of overused terms.

You really think that elevate is less used normally?

Compared to AI generated text, yes. But I'm also going off the physical history of the word 'elevate' in textual history - it had a substantial peak in literature in the 1850s and declined significantly over time.

 

But since it has a significant presence historically - and not so much more recently - it's a statistical outlier and shows up disproportionately in AI generated text as a result because it has no understanding of the change of language, only the summation of all the text fed into it.

 

(I have actually built very small scale AI projects.)

I’ve experimented with ChatGPT a lot, and sometimes still do. It’s a good tool but you can’t let it control you. But I agree, sometimes it’s too obvious that it’s just straight off copied. I did it too in the beginning though. Guilty as heck to that. But it takes away the genuine interaction and integrity.

 

On a side note, I do catch myself often using more vocabulary that chatgpt has. As Arantor said, elevate, empower, etc.

It’s a good tool

I had a first hand look this week at someone using it... shall we say under less than ideal circumstances.

 

My work had a 'Christmas quiz' where the questions were sourced from ChatGPT. More than one of the questions managed to have the answer in the question, and one of the questions about a fictional snowman managed to confuse which particular snowman it was 'thinking' of.

 

I've also had to wrangle with other developers writing code using Copilot and Gemini where what they've got out of it is for much older versions of Laravel than we were using (e.g. Laravel 5.5 versions of code, on a Laravel 10 codebase) which was... frustrating.

 

I haven't actually seen that many cases of it being a good tool, at least on its own, but as part of a larger workflow I could possibly see that... one day.

I had a first hand look this week at someone using it... shall we say under less than ideal circumstances.

 

My work had a 'Christmas quiz' where the questions were sourced from ChatGPT. More than one of the questions managed to have the answer in the question, and one of the questions about a fictional snowman managed to confuse which particular snowman it was 'thinking' of.

 

I've also had to wrangle with other developers writing code using Copilot and Gemini where what they've got out of it is for much older versions of Laravel than we were using (e.g. Laravel 5.5 versions of code, on a Laravel 10 codebase) which was... frustrating.

 

I haven't actually seen that many cases of it being a good tool, at least on its own, but as part of a larger workflow I could possibly see that... one day.

I know. If it’s too complicated it confuses itself. I’ve had a couple of times where I just started cussing it out due to frustration. It does always stay friendly though. No matter what you say to it. 🤣

I will be honest, this is one of the things that bothers me immensely about the collective interactions: the anthropomorphisation of the thing.

 

It doesn't 'stay friendly', because that would imply that it understood what friendly or unfriendly meant, or indeed that it actually understood anything actually said to it, rather than digesting the words down into statistical assessments to align it with statistical outcomes.

 

I know it's tempting to attribute dice to being friendly or unfriendly (especially the b-----d dice that my D&D group of many years ago used to talk about) but there's nothing there behind the curtain.

I beta tested CoPilot. Was not impressed then, am not impressed now.

I’ve experimented with ChatGPT a lot, and sometimes still do. It’s a good tool but you can’t let it control you. But I agree, sometimes it’s too obvious that it’s just straight off copied. I did it too in the beginning though. Guilty as heck to that. But it takes away the genuine interaction and integrity.

 

On a side note, I do catch myself often using more vocabulary that chatgpt has. As Arantor said, elevate, empower, etc.

 

(groan)

 

I guess it's true that the more we read, the more words we'll use. I think teachers expected that to come from good literature and not AI LOL

 

I'm sure you'll come up with better words to substitute very soon. I wonder what the new word for 2025 is going to be that becomes all the rage?

I will use AI to help me with idea generation when I'm out of ideas. That doesn't happen too much.

 

I use AI to build article outlines only to see if it suggests things I should consider adding to my own crafted outline.

 

I use Grammarly AI for editing my written work. I only do it for minor stuff. I try to keep the words I use original; otherwise, who's voice is it actually?

 

When AI suggests an edit due to grammar or whatever, I'll manually do it, too. That helps me get better at it rather than be lazy and press the AI button to correct it.

 

AI can be a great tool, but it can also dumb you down if you let it.

 

I can also tell what's written in AI. I can tell by looking at a person's past content versus what they have been posting lately. The choice of words and the structure of their sentencing and messaging often change.

 

When I detect people doing it, it's almost awkward feeling because it's so obvious.