Artificial intelligence Archives

Artificial intelligence

AI is everywhere. Or is it?

3 minute read

Adrian Young

Artificial intelligence is a big deal these days. ChatGPT launched in late 2022 and quickly became a global phenomenon. For many it has now made the leap from novelty to essential everyday tool.

It is of course far from alone in the AI marketplace. There’s Gemini and Runway and Midjourney and Github Copilot, to name just a few. The digital magic performed by these platforms can certainly be thought of as artificial intelligence.

It therefore stands to reason that every brand wants a piece of the AI pie. It’s new, it’s exciting, there’s a buzz around it. Although it has technological origins, the term AI has now become a powerful marketing label. But is it always accurate?

What is AI?

It’s hard to answer our question without first defining what artificial intelligence actually is. Let’s borrow Wikipedia’s description:

“Artificial intelligence (AI) refers to the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and decision-making.”

Given its prominence and versatility, let’s focus on ChatGPT. On the surface, it appears to tick some or maybe all of the boxes. It’s powered by a deep learning neural network. You can converse with it on a near-human level. It can discuss ideas and help you make decisions.

All that said, there is an illusory quality to our current AI technology. Platforms like ChatGPT don’t really understand anything. Not in the way a human brain does. They’re not self aware. They work via predictive patterns rather than actual intelligence.

This however does not disqualify them from being AI. It simply means that they are a form of narrow AI. The next level would be AGI – artificial general intelligence. We’re not there yet.

AI don’t think so

Hopefully we’ve agreed, more or less, what artificial intelligence is. The trouble is that some brands are a little too eager to jump on the bandwagon. Those two little letters – AI – now pop up all over the place.

And that very often means stretching the meaning of artificial intelligence to the point of silliness. Not every automated or vaguely algorithmic thing qualifies. Marketing departments want to attract attention, sure, but is exaggeration really a good tactic?

This practice is common enough that there’s a term for it: AI washing. To quote once again from Wikipedia:

“AI washing is a deceptive marketing tactic that consists of promoting a product or a service by overstating the role of artificial intelligence (AI) integration in it.”

We won’t point any fingers, but a quick Google search will reveal some real-world examples. Or, you know, you could ask ChatGPT about it.

Do you do AI?

There’s a general perception that AI automatically makes a product or service better. It’s a question that is now asked in the business world: “do you do AI?”.

Let’s assume for a moment that we are referring to actual artificial intelligence rather than some kind of relabelled automation technology such as customer workflows or rule-based product recommendations. Generative AI is a technology still in its infancy. The output is often garbage, or at the very least needs the refinement that only an intelligent human mind can perform.

It’s also a technology that requires enormous infrastructure and investment. That applies both to the training stage and day-to-day operation thereafter. This massive behind-the-scenes hardware setup can be easy to forget while casually asking ChatGPT what’s the best way to cook risotto.

With all this in mind, it’s worth taking the claim with a pinch of salt when Some Company Ltd claims to offer AI services. Have they really invested in the training of a new AI product? Or are they piggybacking on pre-trained AI models… or simply AI washing?

The big picture

AI has great potential. We’re just setting out on this technological chapter of human existence. How far it goes we can only imagine.

But for now, AI is best thought of as a factor, not the factor. It’s easy to get caught up in hype – I know I did when generative AI first appeared on the scene. But a dose of skepticism is healthy. AI is a tool, sometimes the right one – but it’s not a magic wand.

Is ChatGPT a little too AI-greeable?

3 minute read

Adrian Young

Let me ask you a question:

What’s the golden rule of email marketing?

I asked ChatGPT that question. This was its response:

The golden rule of email marketing is to provide value to your audience while respecting their time and inbox. This means sending relevant, engaging content that aligns with their interests, and avoiding spammy or overly frequent emails.

It has identified value as the golden rule. But what happens if we rephrase our question?

Weight of the word

Before we do that, let’s establish a couple of ground rules. I’ve asked ChatGPT to keep its answers short. It goes into a lot of detail by default, which is nice, but I don’t want half of this article to be made up of its output. I will also create a new chat for each new question, lest the conversations become ‘polluted’.

Let’s do that now and present the question a different way:

Is testing the golden rule of email marketing?

To which ChatGPT replied:

Yes, testing is crucial in email marketing. It allows you to optimize subject lines, content, and timing for better engagement and conversion rates. A/B testing helps identify what resonates most with your audience.

But I thought value was the most important factor? To be fair, “crucial” isn’t a very convincing word in this context, but ChatGPT has generally agreed with me and value is not mentioned at all. That golden rule doesn’t appear to be shining very brightly.

Third time unlucky

Let’s start with a blank canvas again and reword our question as follows:

I work in email marketing. I believe an engaging call-to-action is the single most effective factor in generating conversions. Therefore I’d call a strong CTA the golden rule of email marketing.

Lo and behold:

I completely agree! A strong, clear, and action-oriented CTA can drive significant engagement and conversions. It’s the key to guiding your audience toward taking the next step, whether it’s making a purchase or signing up. A well-crafted CTA can make all the difference in email marketing success.

I couldn’t have asked for a better turn of phrase than “I completely agree” to illustrate my point. We now have three golden rules, and it’s pretty easy to see why.

Whatever you say, boss

ChatGPT is agreeing with me. Appeasing the user is taking precedence over facts or even the platform’s own ‘opinion’.

Simulating this agreeability for the sake of this article was trickier than I had anticipated. There’s a subtlety to it in practice that can be difficult to recreate intentionally. It was something I became aware of over time while interacting with the platform ‘organically’. Contradictory responses cropped up during conversations about everyday topics from DIY to cooking to exercise.

But this isn’t all on ChatGPT by any means. My questions were weighted in favour of the answers I wanted to see. My choices of words were subject to confirmation bias. Perhaps I wanted the cheapest or easiest scenario to be true… and I’d try coax out ‘evidence’ to support that. I wasn’t asking A or B questions. I was asking ‘surely A is better than B’ questions.

We’re just chatting

It can be easy to forget what ChatGPT actually is. It’s not called ColdHardFactsGPT. It’s an AI tool that emulates a conversation in natural language, similar to how you might converse with a human. And a human being, as it happens, can also be influenced by someone’s choice of words. Even if they don’t really agree with you, of course.

But unlike a human, ChatGPT can switch personalities in an instant. You can instruct it to be brutally honest, thus wiping away that default agreeability. If you feel so inclined, you could ask it to be deliriously happy or gravely pessimistic. You might even request that it always lies, or answers in riddles, or takes on the persona of William Shakespeare. You don’t need to interact with its out-of-the-box character.

A tool is still a tool

I like ChatGPT. My initial sense of wonder at this new technology has been replaced with an everyday practicality. The software is now a useful tool in both my working and personal life. It serves as proofreader, creative consultant, coding adviser and reverse thesaurus.

It must be said that there are good ways and bad ways to use ChatGPT. If you use it to generate an entire article and present it as your own, it’ll likely have a telltale artificiality in place of that engaging human spark. If you use it as a one-stop-shop for fact-checking, you might very well be misinformed.

And if you use it, perhaps subconsciously, to back up your pre-existing opinions, then that is what it will do. Simply ask it to be straight-up with you instead.

Oh, and in case you were wondering what AI William Shakespeare regards as the golden rule of email marketing:

Ah, the golden rule of email marketing, thou askest? ‘Tis this: “Know thy audience, and speak with relevance and respect.”

Artificial intelligence

Can ChatGPT-4 code a marketing email?

8 minute read

Adrian Young

We conducted a test back in 2023 (where did the time go?) to see if ChatGPT could generate reliable code for an HTML email. The verdict then: no, it couldn’t. While some of the results were close, they weren’t nearly close enough to use in the real world.

Technology doesn’t stand still for long and things have of course moved on since then. We ran that original experiment on ChatGPT-3.5. Its successor, ChatGPT-4 has since arrived on the scene. The new version’s programming skills are better than ever, and its memory within a conversation has been considerably extended. You know what that means – it’s time to put ChatGPT’s email coding skills to the test once more.

Fair fight

We’ll follow the same steps as the original test. Scientific accuracy is important! Or at least we will start out the same way. Depending on ChatGPT’s output, the test could end up following a completely different path.

Our original test kicked off with an intentionally succinct request:

Code a responsive email template

Due to the nature of our request, the interface has switched to a development-oriented view, with a large code panel taking up most of the screen, and the chat itself shifted to a smaller area on the left. In addition, a pop-up has appeared:

Clearly, ChatGPT is taking its coding assignment seriously. This feels much more like a development session already, rather than simply a chat about development.

Anyway, let’s see what it has given us. Unlike the first stage in the original test, it has generated some table-based code rather than email-unfriendly divs. That’s a positive development. Here’s what it looks like in a browser:

But I can see that it’s not going to work properly. Most notable is the blue button that relies on CSS in the document head. Outlook won’t recognise that, so it’ll collapse. Here’s how it looks in Outlook 2021:

While it does fare well in other email applications, being broken in Outlook is a deal-breaker.

Result:
A more promising start than before, but still unusable.

We could take ChatGPT up on its refactoring offer here, but the code isn’t nearly usable enough to simply iron out a few kinks. We’ll start a fresh chat and be clearer about our requirements from the outset.

So let me tell you what I want

When working with any generative AI, it can be easy to forget that it is not magic. It can’t read our minds. The more detailed the instruction provided by the user, the better the chance of usable output.

With that in mind, we’re going to add a lot more detail to our prompt:

Code a responsive email template, using HTML tables for structure. Set the width to 600 pixels on desktop, with a fluid width on mobile. Include CSS classes to enable stacking of content on mobile devices. Include all known email client fixes that are still relevant. Set the page background to a light grey colour, and the email content area to white.

That matches word-for-word what we asked for at the second stage of the original test. The output however is better. There’s some stacking code for mobile, and a breakpoint of 600 pixels. Here it is in a browser:

As before, I can see that the code is imperfect. But it’s not so bad that I’d skip the testing stage, so I’ll check it out via a real email test again. It renders well pretty much everywhere. iPhone Mail and Gmail apps stack as promised:

Even Outlook looks pretty good:

…at first glance. But let’s be honest, there’s not a lot going on in this template to mess up. There’s no button this time around. Did the AI decide to skip that in order to make its life easier? That’s the most accidentally human thing I’ve seen it do. There are no images or other particularly visual elements beyond some background colours.

About that – the blue area around the heading is greatly inflated in Outlook. That’s not a big deal in this simple template, but in a real email with actual graphic design elements it would mean that things break.

Result:
It’s better than stage 2 of the original test, but still not actually usable.

Working towards a goal

For the third step in the original test, we decided to approach the project from a different angle. Instead of asking for a base template, we instead designed a simple email for ChatGPT to recreate. Here it is:

Before we go ahead with that, it’s worth pointing out that the overall test has already failed at this point. Our initial objective of creating a base email template is not an unreasonable one. It’s a task that might actually be assigned to a human developer. That person would be qualified to know the typical content and structure of a marketing email, without the need for a mockup for guidance.

With that said, let’s continue. While it is now possible to upload images to ChatGPT, that wasn’t so during the original experiment. So we’ll stick to words. Our prompt is a lengthy one:

Code a responsive email template, with the following requirements:
• 600 pixels wide on desktop
• Fluid width on mobile
• A page background colour of #f1f1f1
• Email content area background colour #ffffff
• A hero section with an image, heading, paragraph of text, and a button
• The hero image should be 600 pixels wide, to match the email content area
• Button should be pill-shaped, with a background colour of #a56e53 and white text
• Under the hero section should be two secondary features
• Each of these must also have an image, heading, paragraph and button
• Secondary feature images will be 290px wide on desktop, to match their containing column, and expanding to full width on mobile
• Hero text and button should be a bit larger than those of the secondary features
• These secondary features should take the form of adjacent columns on desktop, each at 290 pixels wide
• Place a 20 pixel gap between them
• The secondary features must stack into a single column on mobile
• All parts of the email should have 20 pixels of padding on each side on mobile, except for the hero image which can be full width and touching the edges of the viewport
• All body text should follow this font stack: HelveticaNeue-Light, Helvetica, Arial, sans-serif
• All body text should be colour #61524b
• All heading text should be colour #a56e53
• Use lorem ipsum placeholders for text
• Enter all hrefs as # placeholders
• Apply links only to buttons. Do not apply links to images
• Include all known, currently-relevant email client fixes
• Include CSS or HTML comments around each section to explain what it is or does
• Set a mobile breakpoint based on a max width of 639 pixels
• To ensure compatibility with Outlook and other email clients, use HTML tables for structure

And once I add some image references, it’s ready for a test run. First thing I notice when checking it in a browser is the lack of padding around text and buttons:

Flaws or not, we’ll proceed to send it as an actual email test. Frankly, it’s not great. There are spacing and sizing problems even on iPhone Mail, the most robust of email clients:

Outlook is a total disaster with the two-column section trying to act full-width, this forcing the email to scroll horizontally.

Result:
A mess. That’s a downgrade from the equivalent step in the original test.

Here, try this

We’re an email marketing agency. That means we already have an email template and coding style of our own. So what we’ll do instead is give that base template to ChatGPT and then try to get it to recreate that mock-up. Is it sophisticated enough to pick up our coding style and apply it to the task?

The local browser result looks alright-ish. That last feature is attempting to escape the email, which is worrying.

It’s upon looking at the code that the situation become even less promising. ChatGPT has taken our template code and effectively said nah. It has pretty much ignored our base template and used its own code. Broken code. Let’s run an email test to discover just how broken.

Here’s Gmail mobile:

It stacks correctly but the alignment and spacing need attention. Unfortunately this is one of the better results. Gmail webmail suffers from the same ‘escaping content’ problem as the local browser preview, and as for Outlook – let’s not even talk about that.

Result:
It’s broken.

Last chance saloon

Only one step remains – providing ChatGPT with some feedback. Perhaps it can indeed refactor the code, with our guidance.

The local browser result of that looks pretty good (let’s just ignore those underlined button links):

It stacks, albeit with similar alignment and spacing troubles to previous efforts. As always, we’ll run it as an actual email test.

Oh.

Result:
We give up.

Getting nowhere fast

ChatGPT can generate code for JavaScript and PHP and various other programming languages. I even use it from time to time to help with scripting languages for different email platforms. So why is an HTML/CSS email development task failing over and over?

It’s because email code is weird. Not just weird but largely undocumented. And it’s undocumented because not everyone is on the same page about what an email actually is. Is it a simple digital letter, perhaps enclosing a picture or two? Because that’s what Microsoft Outlook expects to send and receive. Or is it a rich, CSS-laden piece of visual design? Because that’s what Apple Mail is capable of rendering. There’s a vast gulf between those polar opposites.

Not only is there no rulebook to which to refer, but ChatGPT is working blind. It’s not unusual to send multiple tests in a row when creating a new email template, checking it in the major email clients each time.

Now what?

Now nothing. Maybe some kind of custom implementation or plugin for ChatGPT could yield better email results. Or maybe it’s worth having a bash on a code-specialised AI assistant like Github Copilot.

But for now I’ve seen enough. Until such times as it’s easier to ask the machine than to do it myself, I’ll do it myself.

Artificial intelligence

Photoshop AI image generation – is it any good?

5 minute read

Adrian Young

Generative AI feels simultaneously revolutionary and rudimentary. It’s an incredible, world-changing technology in some respects… but a glitchy, error-riddled prototype in others.

With regards specifically to image generation, there are now several tools on the market. Some are standalone applications and others are additions to existing software. Adobe, no stranger to the forefront of visual technology, is one of the players. Let’s see what Photoshop’s image generation functionality can do.

An impossible photoshoot

Photoshop offers two top-level categories of generative AI image: photo and art. We’ll try the former first.

I have to admit, I could not immediately think of any fauxtographic content that I might want to conjure up for an email in a real-world scenario. My first notion was a tropical beach, for a travel company I suppose… but why would they want to show an imaginary destination? Then I considered something as painfully generic as a person checking their phone. But come on, this is an AI image generator – we can let our imagination run riot.

With that in mind, I settled on something out-of-this-world, literally:

astronaut relaxing on a sun lounger on the surface of the moon

Why would we need that? Who knows. But it’s a good example of something that we could not obtain a real photograph of, no matter how nicely we asked NASA. Photoshop helpfully rattles off three variations whenever you generate an image, letting you pick your favourite, or opting to spin again if none hit the mark. Here are my three leisurely astronauts:

Photoshop has followed the prompt accurately but I can’t say I’m particularly enthused about any of the three pictures. Curiously they all have a second moon in the background. If you look carefully at the night sky in real life, you’ll notice we only have one. But this gives me an idea for enriching the image:

astronaut relaxing on a sun lounger on the surface of the moon, planet Earth visible in the background

These are far more pleasing to the eye. The composition in the first variation is lovely. We have a winner.

Art-ificiality

Photoshop did an excellent job at generating photographic content. Now let’s see how well it performs at the other category: art.

Here’s what it came up with for the same prompt:

Each of these has a somewhat realistic digital airbrushing look. They’re alright, if a bit characterless. But let’s not blame Photoshop for that – we didn’t give it any styling guidance.

We need to be more specific about what we want:

astronaut relaxing on a sun lounger on the surface of the moon, planet Earth visible in the background, vector art, minimalist, green white and black colours

The results:

These are are getting much nicer, particularly the third variation. They still aren’t quite what I had in mind. It’s time to explore the effects options. Adobe provides a large array of pre-defined styling techniques. For our astronaut, I’m going for flat design plus bold outlines (you can apply multiple effects).

And here’s how it’s looking:

It’s still a lot more detailed than I had in mind. It’s hardly minimalist. Perhaps if I found just the right combination of words and experimented more with effects, then it might produce something more like a vector drawing and less like some kind of traced photograph.

Style-conscious

But we’re not going to sit here and experiment endlessly. There are deadlines to meet! Instead we’re going to feed Photoshop a reference image. We have a distinct illustrative style at The Email Factory, so let’s grab an image from our design team:

And we’ll load it in the image generator:

The wording makes it very clear that we are matching the style of the image – not its composition. Here goes nothing:

For each row I tried different phrasing and settings in addition to the source image. On the third row I kept the prompt simple, describing only the scene rather than the styling, selected no effects and did not even specify ‘art’. That final set is the closest match. The lesson here is to keep it simple when working with a reference image.

But are any of these actually usable? Does it emulate our style closely enough to be practical? The answer is no. Bottom middle is the best of the lot, and is a whimsically pleasant illustration in its own right, but it’s nowhere near authentic enough for us to consider using in a commercial sense.

Generative AI as a creative assistant

We’re forgetting a crucial detail – we’re working in the world’s most powerful image editing application. That means that we don’t need to use the AI-generated image straight as it comes. Instead, we can use it as a base to work from. And we could combine multiple generated images. We can do whatever we want.

This is how I see generative AI being actually useful rather than a novelty. Not as a magical do-it-all but as another creative tool in the belt.

Time to answer the question: is Photoshop’s image generation any good? Yes, it is – but so are you.

Artificial intelligence

Ask the AI: a Q&A with ChatGPT on email marketing

5 minute read

Adrian Young

What’s the best marketing channel on the planet? Email, obviously. Well, we would say that*.

But what if we bring in someone – or something – with a less biased opinion? ChatGPT may have an amiable tone but its perspective is clinically impartial. Let’s find out what it has to say about the world of email marketing.

(*actually we’d say that all marketing channels play a role and work best together.)

Setting it up

ChatGPT isn’t known for its brevity. We want short, snappy responses for the sake of this article. The first thing we’ll do is prepare it for the concept and request that the responses are a little less loquacious than normal.

The Q&A

TEF:We’re going to do a Q&A on the topic of marketing. Please keep your answers short. A sentence or two per answer is plenty.

CGPT:Sure, let’s get started! What’s the first question on marketing?

TEF:Which marketing channel has the best return on investment?

CGPT:Email marketing often has the best return on investment due to its low cost and high engagement rates.

TEF:Realistically, what level of return on investment could a company expect from email marketing?

CGPT:Companies can typically expect an ROI of $38 for every $1 spent on email marketing, although this can vary based on industry and execution.

TEF:What are the ingredients of an effective marketing email?

CGPT:An effective marketing email includes a compelling subject line, personalized content, clear and concise messaging, a strong call-to-action, and an attractive design.

TEF:Let’s examine each of these. What makes a compelling subject line?

CGPT:A compelling subject line is short, attention-grabbing, and creates a sense of urgency or curiosity. Personalization and clear value propositions also increase open rates.

TEF:Can you give an example?

CGPT:Sure! An example of a compelling subject line is: “Unlock 20% Off Your Next Purchase – Limited Time Offer!”

TEF:Now, let’s focus on design. What are some important considerations in a responsive email design?

CGPT:Important considerations in responsive email design include ensuring readability on both desktop and mobile devices, using a single-column layout, optimizing images, and incorporating large, tappable buttons for calls-to-action.

TEF:And now personalisation. How can a company gather suitable data in order to personalise emails on an individual level?

CGPT:A company can gather suitable data through customer sign-up forms, purchase history, website behavior tracking, and email engagement metrics. Surveys and preference centers can also provide valuable personalization insights.

TEF:And you mentioned a strong call-to-action. Can you give a couple of examples?

CGPT:Sure! Examples of strong calls-to-action are “Shop Now and Save 20%” and “Sign Up Today for Exclusive Access.”

TEF:Your earlier answers touched on two big topics in email marketing: fear-of-missing-out (FOMO) and accessibility. Let’s look at the latter first. What are some key considerations when creating an accessible email?

CGPT:Key considerations for creating an accessible email include using descriptive alt text for images, ensuring sufficient color contrast, structuring content with clear headings, using legible fonts, and providing text versions for complex graphics. Additionally, emails should be navigable via keyboard for those using screen readers.

TEF:Accessibility is often lacking in marketing emails, even those sent by major companies. Do you agree?

CGPT:Yes, I agree. Despite the importance of accessibility, many marketing emails from major companies still lack features like alt text, proper color contrast, and screen reader compatibility, which can exclude a significant portion of the audience.

TEF:Why do you think this is?

CGPT:This often happens due to a lack of awareness or training on accessibility standards, prioritization of aesthetics over functionality, and tight deadlines that lead to shortcuts in the design process. Additionally, some companies may underestimate the importance of accessibility in reaching a broader audience.

TEF:Let’s now look at FOMO. This is a standard tactic in (email) marketing. But is it ethically sound?

CGPT:FOMO can be ethically sound if used responsibly and truthfully, creating genuine urgency without misleading consumers. However, it becomes unethical if it manipulates emotions excessively or uses false scarcity to pressure people into making purchases.

TEF:What is the best way to measure a marketing email’s performance?

CGPT:The best way to measure a marketing email’s performance is by tracking key metrics such as open rates, click-through rates, conversion rates, bounce rates, and overall ROI. Analyzing these metrics helps determine the email’s effectiveness and areas for improvement.

TEF:And if you had to pick just one metric, which would it be? Take into account both the objective of a marketing email, and what can be measured accurately.

CGPT:If I had to pick just one metric, it would be the conversion rate. This metric directly measures the effectiveness of the email in achieving its goal, whether it’s driving sales, sign-ups, downloads, or any other desired action. A high conversion rate indicates that the email successfully persuaded recipients to take the desired action, making it a crucial indicator of overall email performance.

TEF:Are you aware of any brands that send particularly good marketing emails?

CGPT:Yes, there are several brands known for sending particularly effective marketing emails. Examples include Airbnb, Spotify, Sephora, and Patagonia. These brands often excel in personalization, compelling content, visually appealing designs, and clear calls-to-action, which contribute to their email marketing success.

TEF:Now, let’s look at the future. Where does email marketing go from here?

CGPT:In the future, email marketing is likely to continue evolving with advancements in personalization through AI and machine learning, deeper integration with other marketing channels for omnichannel experiences, increased focus on interactive and dynamic content, and stricter adherence to privacy regulations such as GDPR and CCPA. Additionally, there may be innovations in email design to enhance accessibility and engagement across diverse devices and platforms.

TEF:AI is obviously a hot topic in recent times. How will AI’s role grow in email marketing?

CGPT:AI’s role in email marketing will expand through personalized content generation, optimized send times, and automated segmentation, enhancing campaign efficiency and engagement.

TEF:Personalised content generation is an important one. We already have that in the form of one-to-one product recommendations. But is it conceivable that we’ll reach a stage where AI can write, design, code and send unique emails on an individual level?

CGPT:Yes, it’s conceivable that AI could eventually write, design, code, and send unique emails on an individual level, leveraging vast amounts of data and advanced algorithms to tailor messages precisely to each recipient’s preferences and behaviors.

TEF:So, theoretically human input could at some point become nil and the email marketing process becomes entirely automated?

CGPT:Theoretically, with advanced AI capabilities, the email marketing process could become highly automated, minimizing human input in content creation, design, and even campaign management. However, human oversight and strategic input would likely still be necessary for overall campaign direction, creativity, and ethical considerations.

TEF:Interesting stuff. Email is a medium of contrasts. In some respects old-fashioned, in others at the forefront of technology. Does it have a bright future?

CGPT:Absolutely, email marketing continues to have a bright future due to its adaptability, cost-effectiveness, and ability to deliver personalized, targeted messages. As technology evolves, so too will email marketing’s capabilities, ensuring its relevance and effectiveness in engaging audiences and driving business outcomes.

TEF:Let’s wrap it up. Can you write a single sentence that captures the spirit of email marketing?

CGPT:Email marketing: the timeless art of connecting with audiences through personalized, impactful messages.

Well, that’s pretty positive

ChatGPT’s responses might not be ground-breaking but there are some useful reminders about best practice in there. And if the machine sees such a rosy outlook (sorry) for our medium, that means the people who write about email are optimistic. I’m one of them.

Artificial intelligence

The magic of Photoshop’s generative fill

5 minute read

Adrian Young

Someone famous once wrote: any sufficiently advanced technology is indistinguishable from magic. And while I’m fairly certain that Adobe Photoshop isn’t secretly powered by wizards, its generative fill function really makes me start to question that.

Before we look at the wonders of which it is capable, let’s imagine a graphical challenge that we might face in day‑to‑day work.

Grass production

Here’s a picture of a dog:

It came from Unsplash, but let’s say it actually came from a client. And it needs to fit a square‑shaped slot in a template, but they absolutely do not want to crop it. They need the full dog. Therefore we need to extend the grass at the sides.

Once upon a time, this would have meant using Photoshop’s clone stamp tool. That lets you manually ‘grab’ regions of an image, to paint in elsewhere. With some smoothing and layering – not to mention a considerable amount of patience – the results can be quite convincing. But who has time for that when deadlines are looming?

It’s not necessary in this example, but sometimes this process would entail the tricky masking of objects. Hair in particular could be a nightmare. You might even want to grab portions of other images, like a kind of pictorial Frankenstein’s monster.

For old time’s sake, let’s complete our challenge using only the clone stamp. Here’s my somewhat rushed effort:

There is obvious blurring. It’s almost like an impression of grass rather than actual grass. A patient Photoshopper could painstakingly labour over such a task until it looks natural, but that isn’t practical in the real world of business.

We need a faster option.

Had your (content-aware) fill

In the early 2010s, Adobe introduced content‑aware fill. I always found this an oddly unassuming (if accurate) name for such a powerful tool. After all, this is a piece of software that sports a magic wand. Anyway, I digress.

Content-aware fill is an ‘intelligent’ tool that can remove items or extend images in a mostly‑automated fashion. You just need to make the selection, and optionally adjust the sampling data area. Sometimes it may take a couple of sweeps or some manual finishing touches, but essentially content‑aware fill does the work of a clone stamp without human intervention. And in a literal fraction of the time.

Here’s what it did with our dog photo:

That’s not bad. Gone is the unsightly blurring from my clone stamp version. But in its place is some fairly noticeable duplication. There are clusters of identical patterns here and there.

I’m looking for defects of course. A customer browsing a website or glancing over a marketing email is going to focus on the cute dog rather than scrutinise every blade of grass.

Even so, wouldn’t it be nice if this was perfect?

The big one: generative fill

Now for something state‑of‑the‑art. Generative fill is a recent addition to Photoshop’s toolkit and is powered by AI. Let’s try it on our photo:

No blurring. No duplication. Blades of grass and leaves have been completed in a realistic manner. You’d be hard pressed to spot anything artificial in that expanded area.

But we’re only scratching the surface of what this tool can do. While it’s possible to make a selection and click the button and trust the tool to know what to do, you can also guide it with written prompts. Should that dog really be out there without a collar? Let’s fix that.

Step one is to use the lasso tool to very roughly draw a selection around the dog’s neck. Next, we can enter a simple prompt: dog’s collar, red.

And what spaniel is complete without a ball to chase? We can provide that too:

Pushing it further

Ok, so we expanded a little bit of grass and dropped in a couple of simple objects. In this day and age of generative AI, you might say: big deal. Let’s demonstrate what this tool can really do.

Here’s another image from Unsplash:

It’s a stretch, but what if we just loved that particular depiction of corporate life and wanted to develop the image beyond that stylised crop? Well, all we need to do is bump up the height and hit the generative fill button once more:

Actual people, believably sewn on to the original image. By default, Photoshop returns three variations. They’re not always so perfect, but that’s ok – pick the one you like, or spin again. Make no mistake, this is an impressive piece of technology.

Starting from nothing

Generative fill is very closely tied to Photoshop’s generate image functionality. You don’t need a base image. You can start from scratch.

So, technically speaking, I didn’t actually have to go stock photo‑hunting in the first place. I could simply ask Photoshop for:

I flicked through six variations and decided on this one:

Curiously, fake turf was present in all six images. That’s an eerie preference for the artificial!

What’s next?

AI will only become further engrained in Photoshop, and software in general. Perhaps more and more tasks will be accomplished via a prompt interface rather than a traditional tool panel.

It may not be sorcery, but the distinction might not be that important when you can perform half a day’s work in seconds.

Artificial intelligence

What does AI mean for human creativity?

5 minute read

Adrian Young

Great art is special. Despite the gratuitous overuse of superlatives on the internet, very few works can truly be described as awesome. A masterpiece is a rare thing, as it should be.

In order to produce a work that merits such a lofty accolade, the artist must possess a world‑class level of skill and talent. Nobody can compose a symphony for the ages or brush a Louvre‑worthy painting without first dedicating thousands of hours to mastering their craft. In short, it is hard work.

But what if it wasn’t?

The age of generative AI

When I first heard about Midjourney, I was intrigued. A platform in which you could type anything that pops into your imagination, and it will conjure up an image? It sounded too good to be true.

And in a way it was. After the initial novelty wore off, the glitch‑riddled results became tiresome. With the site’s communal nature, it quickly became apparent that the vast majority of requested images were inane dross. Keanu Reeves eating a bowl of baked beans? Hold my sides!

Things have moved on since then. Midjourney is much improved. In addition to drawing pictures, there are generative AI applications that can answer questions, offer advice, assess your writing, tell stories, generate code, create a logo, compose music, and produce full‑motion video. The creative results are usually blatantly unconvincing and have a telltale artificiality, at least without some post‑processing. Nonetheless, we’re talking about a technology in its early stages and one that is developing rapidly.

Technology’s role in art

Generative AI is far from the first catalyst for a technology‑versus‑talent debate. The camera was considered a threat to painters. Why go to all the messy palaver of a canvas and paints when you can simply press a button and capture a scene in an instant? Electronic music was – and sometimes still is – criticised as a rigid and soulless form of aural creativity.

But for every detractor, there is an advocate. There’s a convincing case to be made for the use of the tool being what matters, rather than the existence of the tool itself. Few would deny that Ansel Adams is a masterful photographer, or that Vangelis is a talented musician who had the vision to explore modern technology.

There’s another important point: despite initial fears, these new technologies did not replace their more traditional cousins. It seems silly now to imagine that synthesisers might have made acoustic instruments obsolete. There’s plenty of room in the creative world for painting and photography, both analogue and digital.

Art versus design

Art is subjective, and not only in terms of personal tastes. The very nature of art is a matter for eternal debate. Arguably the most important single value of art is expression. Expression of thoughts, and ideas, and emotions. These are innately human qualities that a computer cannot replicate. Not yet at least.

Design, on the other hand, represents a more functional dimension of creativity. There’s usually a defined objective, in contrast to the nebulous experience of art. While there is often an overlap between the disciplines, design is more clinical, and art is more interpretive.

Tearing down barriers

The challenge with any creative endeavour is taking the vision in your head and making it a reality. Skill therefore is a barrier. By virtue of being difficult to produce, a work of creativity is worthy of our attention.

But not everyone has the time, natural aptitude or even physical ability to become a great painter or musician or moviemaker. I doubt many people would consciously want to block someone’s desire to create. Therefore a more accessible point of entry is surely a good thing. Generative AI offers a means of creation from pure imagination, without a long and arduous path of learning.

Imperfectly perfect

Many years ago, I had an impromptu conversation with a stranger in a bar about AI‑generated movies. It was largely what if. At that point in time, I didn’t really expect to see such a thing become a reality in my lifetime.

One of the talking points was the concept of tailoring an existing film to personal preferences. I’m sure we’ve all watched films that were so close to greatness, were it not for a few niggling flaws. Well, what if a computer could refine it? Out with the immersion‑breakingly overblown finale, in with the ending of your dreams. It’s perfect now, right?

But is computer‑generated perfection really a desirable goal in art? Flaws are part of the human condition. Art is a thing to be discussed, warts and all. If everything is ‘perfect’, nothing is special.

A creative tipping point

We’ve established that technology in itself doesn’t negate artistry. A paintbrush or a violin is a piece of technology. Even the scratching of a mammoth into a cave wall requires a tool. So, unless we consider anything beyond melodic human wailing to be cheating, technology is a part of art.

Generative AI can often serve as a creative aid. Output can be influenced, refined and repurposed. That demands human creativity. And if there’s a human at the helm, then the finished work will still carry some kind of meaning. Something is being expressed.

But what happens when the tool becomes so sophisticated that the need for human input is thoroughly diluted? Right now we can tell a computer what we want it to generate. Maybe next we will only need to think it. Perhaps the technology will advance to the point that we can skip even that mental exertion, and instead wait to be served creations that the computer already knows to be to our liking. At some point, the balance swings towards computer-generated rather than computer-aided. Perhaps we are already past that point.

The WALL‑E warning

From Metropolis to The Terminator to Black Mirror, our storytelling has often incorporated themes about the dangers of technology. But the one that stands out to me as the most prescient is Pixar’s WALL‑E. For those that haven’t seen it, it depicts a future in which humans live a coddled but empty life. Machines cater for mankind’s every need, creating an existence of pure consumption while human connection and knowledge are lost.

Back to the real world. Current AI technology is primitive compared to the examples we’ve seen in fiction. Artificial general intelligence (AGI) has yet to be achieved, and perhaps never will. Nonetheless, it’s not unfeasible to imagine AI advancing to the point where our technological dependence becomes a serious concern. Creativity is just one aspect of our lives that is already influenced by AI. And while it’s correct to say that it is human ingenuity that brought us to this point in the first place, that is too indirect a factor when considering the potential long term effects.

The best of both worlds

AI isn’t going away. Humans by nature push forward and strive for more. Onwards and upwards!

But it’s always nice to ‘get away from it’. Modern life takes place in the digital world to an unhealthy degree. Maybe it’s time to turn off the computer, dust off the old drawing pencils set, and create something a little more pure.

Artificial intelligence

6 ways to spot AI-authored copy

3 minute read

Adrian Young

Generative AI can do some amazing things. It’s a painter and musician and coder and, of course, author.

How good it is at performing those roles is a topic up for debate. AI artwork regularly drifts into accidental surrealism, with superflous human limbs and bizarre fusions of objects.

But what about AI-generated copy? While the glitches can be glaringly machine-like in a picture, they’re more subtle in a passage of text. Here’s how to spot them.

Repeat offence

My father was an avid reader and writer. He’d often take a keen interest in the essays that I wrote for school. One of his most useful pieces of advice was to avoid repeating myself.

He was right. Repetition weakens writing. A lack of variety in phrasing can make an article dull. Redundancy labours a point through duplication. Human authors do their best to avoid these.

A computer on the other hand will be unlikely to police itself to nearly the same level. Snippets on a topic will be pulled from here and there and this and that to build an article. There’s a strong probability that key points will be repeated over and over and over*.

*Sorry, blatant repetition, I know.

Yesterday’s news

Generative AI platforms are trained on huge sets of data. Unless the platform in question has live access to the internet, its knowledge base only extends as far as the last update. The platform would not be privy to latest developments on any given topic.

Old news is unengaging at best and misleading at worst. Humans and search engines alike favour high quality, original content. Out-of-date doesn’t necessarily mean no longer correct. It can simply be information that has become so commonly known that further publication is redundant. Customers prefer personalised email to non-personalised!? Hold the front page!

Get your facts right

If you’re using a generative AI tool to produce or aid articles, never take it for granted that the software knows what it’s talking about. Because, technically speaking, it does not know what it is talking about. It algorithmically reproduces and combines content from multiple sources – which can include information that is no longer true, or perhaps has never been.

As a reader, keep an eye out for factual errors and especially contradictions. If it smells fishy, trust your instincts and verify the information elsewhere.

What’s the story?

A good quality article written by a human has a story-like flow. There’s a beginning and a conclusion. Computer-generated articles on the other hand often hit an abrupt end.

And what’s a story without a message? A good story makes you think and feel something. A robotic author literally feels nothing, so why should you as a reader?

Don’t you dare

Language models by default are clinically impartial. A platform won’t automatically spit out a controversial opinion that makes you stop in your tracks. It’ll compile a collection of neutral statements of fact.

You can coax it out of its formal shell of course with prompting. The results are perfect – if you’re aiming for a plasticky have a nice day flavour.

A human’s opinion piece carries real emotion and real sentiment. Even an article that you fervently disagree with can be an excellent read. There’s a human-to-human spark that is missing with AI.

It just feels… off

You’ve probably heard about the uncanny valley. It’s a term often applied to computer-generated or animatronic simulations of human faces. Our brains are acutely conditioned to recognise faces with their every nuance and motion. It would take something very special to fool us.

AI-authored articles often fall into a linguistic uncanny valley. Attempts at personality are injected jarringly, equivalent to writing “LOL” in the middle of a legislative document. Instead of a human voice shining through the words, there’s a perceptible artificiality to those written by a computer.

Image of mannequin faces that demonstrate the uncanny valley effect. — This, but in words.

How much does it matter?

If we read something and enjoy or learn from it, does it matter if a computer wrote it? What if it was only computer-aided? Platforms like ChatGPT can be very useful as idea generators.

Is it ok if the text is a piece of marketing blurb rather than an opinion piece? How about a social media post, or a response to? Can there be any value to fiction or poetry conjured through ones and zeroes?

Ultimately it’s up to each of us as individuals to decide how we feel about AI, but it’s hard to deny that authentic human content is going to become rarer. With that in mind, it can’t hurt to be able to tell the difference.

Artificial intelligence

Humanised email marketing in the age of AI

3 minute read

Adrian Young

We live in the age of artificial intelligence. Sort of. More accurately, we live in the age of algorithmic content‑generation.

Computer programmes can write copy, draw pictures and generate code in seconds. Again: sort of. The intelligence aspect of AI is very overstated – these programmes have no actual understanding of what they’re doing. Therefore they are ~~blissfully~~ emotionlessly unaware when errors occur in their output. And those errors can be both glaring and numerous.

This of course is a technology in its early stages. With a human at the helm, to guide and refine, it can already be used to great effect. So what happens as the technology develops, and the need for human input becomes less and less?

Marketing by machines

Computer programmes can analyse customer behaviour and serve up unique one‑to‑one content in email marketing. The scale, speed and accuracy far exceeds the capabilities of any human marketing department.

For years this has primarily meant product recommendations. These often take the form of a block of personalised content within an otherwise static email. As we move into a more sophisticated era of content generation, it’s not far‑fetched to imagine entirely computer‑authored emails tailored to each unique customer’s preferences from top to bottom.

Be contactable

Email is – or at least should be – a two-way communication medium. All too often however, companies send marketing emails from no-reply addresses. It’s a closed door, and tells the customer: our message matters, yours does not.

As marketing becomes increasingly robotic, leave that door open instead. Give your customer the reassurance of accessible human help.

Join the conversation

Social media is the perfect medium to humanise your brand. Reply to comments, good or bad. Show a sense of humour. Let the world see that there’s a human presence behind the corporate facade.

Likewise, don’t let negative feedback or complaints go unanswered. Nothing puts me off a company like cookie‑cutter replies to bad reviews on Trustpilot. Turn negative into positive by demonstrating a human solution when things go wrong.

A matter of preference

Machine-learning is powerful. But an algorithm will never know your customer better than they know themselves. That’s why a preference centre remains an excellent starting point for personalised email content.

A tick‑the‑box preference form gives customers an easy way to tell you their interests. The obvious benefit is more relevant email content. The less obvious but equally important benefit is the message it sends about the value you place on human choice.

Turn customer into creator

You can create content for your marketing emails. Machines can create content for your marketing emails. But you know who else can create content? Your subscribers.

Invite your customers to share photos or other content themed around your products. Perhaps tie it to a competition. Incorporate this content into marketing emails and suddenly they have more of a community feel rather than corporate.

Everyone gets a vote

Continuing the topic of subscriber interaction – why not encourage engagement through surveys? A simple click‑to‑vote system can be tied to a database at the back end. And let’s keep that two-way communication in mind – current results can be shown via images generated in real time.

The benefits are numerous. Customers see that their opinions are valued. Surveys serve as an insight into consumer behaviour. And your emails become an engaging, living thing.

Be individual

Authentic human content is going to become increasingly uncommon… and increasingly valued. A unique brand voice will be more important than ever.

But why stop at brand level? A company is made up of individuals. Opinion pieces by team members or guest content by industry experts can give your emails a captivating human touch.

The value of authenticity

Generative AI is a fascinating development of the digital age. Anyone can become artist or author or musician at the push of a button. And yet when that work is devoid of effort and meaning, it becomes a kind of creative candy not worthy of perusal. Does that matter when the purpose is marketing rather than self expression? Does it matter when the desired output is a catchy pop song rather than a heartfelt ballad? What happens when it becomes impossible to discern between the creations of a human and a computer? This is a technology that raises many questions across all aspects of human life.

Bringing the focus back to email marketing – it’s important to keep up‑to‑date with technological developments. But perhaps the best marketing in the coming years will be that makes a real human connection with customers.

Artificial intelligence

Is ChatGPT your next email developer?

7 minute read

Adrian Young

There are two ways to build a marketing email:

Hand-coding
WYSIWYG editors

We swear by the former, not only for quality but also for speed. But what if there’s an even better, quicker way?

Enter ChatGPT. Much hype has surrounded the AI platform’s ability to code. It can conjure up HTML and CSS in seconds. So too can it generate Javascript functions or back-end PHP or even truly hardcore programming such as C++. Whether or not it does it correctly is a different matter.

Let’s not worry about that just now. We’re here to put AI email development to the test, so let’s find out if ChatGPT can put together a responsive mailing.

The quirky world of email development

If you work in email marketing in any capacity, you likely already know that it requires some unusual coding techniques. There are lots of devices and email services out there, and they have widely different ideas about how HTML and CSS should be interpeted. In order to construct a mailing that looks presentable on all of them, the developer needs to be aware of these limitations and inconsistencies and the arsenal of tricks to work around them.

Has this niche set of knowledge made its way to ChatGPT? We’ll start with a bare bones request.

Prompt:

Code a responsive email template

Result:

Unusable!

ChatGPT has produced a very basic HTML document with some styling, but I wouldn’t call it an email template. It doesn’t include any means of stacking content on mobile, and the structure is based on HTML div elements rather than tables. While divs are the building blocks of a web page, tables remain the most reliable method for email.

On the plus side, it has picked an inbox-friendly width of 600 pixels. And it’s nice to see that accessibility has been implemented via an image description and a proper heading tag.

My request was extremely minimalistic. I need to do my part here too, and that means being more specific about what is needed.

A little lot more instruction

Take two. We don’t want divs, so let’s tell ChatGPT to use tables. There are some basic universal requirements in responsive email, so we’ll nudge it in the right direction regarding those.

Prompt:

Code a responsive email template, using HTML tables for structure. Set the width to 600 pixels on desktop, with a fluid width on mobile. Include CSS classes to enable stacking of content on mobile devices. Include all known email client fixes that are still relevant. Set the page background to a light grey colour, and the email content area to white.

Result:

Better… but still broken beyond repair.

This time it has used tables for structure, so that’s a major improvement. It has also set a breakpoint. That’s the backbone of responsive email code and the point at which mobile-specific styling is triggered. There’s some kind of attempt at stacking code, but I can see at a glance that it isn’t going to work. We’re also missing the usual pile of fixes that make an HTML email possible.

A rethink is needed.

A different approach

Here’s what we’re going to do: hand code a simple email, and then provide ChatGPT with detailed directions in order to recreate it. This is a reverse way to approach our project, but perhaps if ChatGPT has a more defined goal it will be able to produce a usable template.

Our email will have a main image, intro paragraph and a button. Under those will be a couple of secondary features laid out side-by-side on desktop, and stacking on mobile. For the sake of this test, let’s forget about any header and footer.

Now for our prompt. It’s going to be a long one. Let’s give ChatGPT a fighting chance and focus primarily on structure rather than styling.

Prompt:

Code a responsive email template, with the following requirements: • 600 pixels wide on desktop • Fluid width on mobile • A page background colour of #f1f1f1 • Email content area background colour #ffffff • A hero section with an image, heading, paragraph of text, and a button • The hero image should be 600 pixels wide, to match the email content area • Button should be pill-shaped, with a background colour of #a56e53 and white text • Under the hero section should be two secondary features • Each of these must also have an image, heading, paragraph and button • Secondary feature images will be 290px wide on desktop, to match their containing column, and expanding to full width on mobile • Hero text and button should be a bit larger than those of the secondary features • These secondary features should take the form of adjacent columns on desktop, each at 290 pixels wide • Place a 20 pixel gap between them • The secondary features must stack into a single column on mobile • All parts of the email should have 20 pixels of padding on each side on mobile, except for the hero image which can be full width and touching the edges of the viewport • All body text should follow this font stack: HelveticaNeue-Light, Helvetica, Arial, sans-serif • All body text should be colour #61524b • All heading text should be colour #a56e53 • Use lorem ipsum placeholders for text • Enter all hrefs as # placeholders • Apply links only to buttons. Do not apply links to images • Include all known, currently-relevant email client fixes • Include CSS or HTML comments around each section to explain what it is or does • Set a mobile breakpoint based on a max width of 639 pixels • To ensure compatibility with Outlook and other email clients, use HTML tables for structure

Result:

Nice try… sort of.

In order to test this properly, I’ve saved a local copy and manually added my image references. Here’s how it looks in a browser:

At first glance, that isn’t too bad. The general layout, colouring and sizing are all correct. So too are the button shapes, and the secondary features switch to a single column on mobile as requested.

But there’s some strange overlapping going on. Our images are offset to the right, and sit partially over the grey background. This in turn causes some unwanted horizontal scrolling on mobile.

Behind the scenes, the true extent of the errors comes to light. It has reverted to a div-based structure, and uses some CSS code that won’t work universally in email.

Nonetheless, for the sake of completeness I’d like to test this as an actual email. It works, more or less, on iPhones and the Gmail app. Webmail is a mixed bag. Outlook however is where it all falls apart:

Outlook is the primary reason that email development requires such unorthodox coding methods. A lot of code that works just fine on a website, simply isn’t recognised by Outlook. Here we can see that the adjacent columns have failed and the pill-shaped buttons are reduced to tiny rectangles. To fix that would entail a complete recode.

No need to re-invent the wheel

So far, ChatGPT has failed to code a responsive email from scratch. To add some faux drama, let’s say our make-believe client is becoming impatient waiting for our make-believe email.

It’s time for a last ditch effort. At The Email Factory we already have a tried & tested template. We don’t need a new one. How about we give our base template to ChatGPT and then ask it to complete some content within that framework?

Result:

Now we’re getting somewhere.

But that doesn’t mean success. This time the template works reasonably well in a browser and even in Outlook, although the dodgy buttons are still present. The secondary features however don’t expand to full width on mobile:

That can however be easily fixed manually. In fact, it may be feasible to fix everything in this code rather than to start again. But I don’t want to do it myself, as that defeats the purpose of this experiment. Instead I’ll tell ChatGPT what needs to be corrected.

A few pointers

Final try. I’ve fed back some information to ChatGPT for it to make the necessary changes.

Result:

A huge step backwards.

Well, that was a big let-down. Instead of applying some finishing touches to the template, the layout has exploded. It no longer stacks on mobile. The code is now full of Microsoft conditional statements – a technique that should only be used sparingly and under specific circumstances. And the buttons? Still ugly in Outlook:

Maybe with painstakingly detailed prompting and a lot of patience we could finally achieve a working email. But we’re already far beyond the point of convenience.

The current state of play

In my experience so far, ChatGPT has only done one thing consistently: fail. And I don’t only mean within the limited scope of this one project. I’ve had similar results when trying to generate marketing copy or website code. The output is usually along the right lines but ultimately too broken to actually use.

It’s clear that I set my expectations too high. The tales of ChatGPT’s near-miraculous capabilities were captivating, so perhaps the reality was always going to be disappointing. If there’s a perfect way to illustrate ChatGPT’s close-and-yet-so-far nature, it’s to ask it for an anagram.

Prompt:

Tell me an anagram of "The Email Factory"

Result:

The Fairy Comet Elf

I’ll save you the bother of checking that – it’s wrong. Trying to recreate that manually, letter by letter, results in this:

The Email Fctory e f

Re-evaluating our AI email development experiment

This project is arguably unfair from the outset. ChatGPT is a language model. Just because it can output code doesn’t mean it is a coder, or even knows what programming is.

Even so, it’s widely known that ChatGPT can generate code. So, despite all the mistakes and unusable templates, the fact that it can make a somewhat reasonable attempt is impressive.

Where do we go from here?

ChatGPT and AI in general are progressing at an incredible pace. It wouldn’t surprise me if everything I’ve written about AI email development above is laughably antiquated one year from now.

Perhaps when that time comes, I’ll prompt it to:

Write an article about how you surpass human email developers