Welcome back to our monthly blog series where I, your humble Senior Community Evangelist, recap our latest Litmus Talks and note key takeaways (and offer a few thoughts of my own) to occupy your precious brainspace.
This month, I’m sharing my thoughts on our dive into Artificial Intelligence (AI) marketing services in “Do Email Marketers Need AI?” Hosted by… well, I suppose I was hosting this one, wasn’t I? I was joined by Matt Byrd, creator of the AI subject line tool, Subjectgeni.us and Head of Growth at Homepace.
Read on for highlights and key takeaways (or jump ahead!):
|Table of contents
What do we mean by “AI”?
AI is the use of computers, machine learning and algorithms to perform tasks that previously could only have been accomplished by human intelligence.
AI tools fall into two broad categories: machine learning and generative production.
- Machine learning is processing a mass of data to make predictive models (think recommendation engines) that identify patterns or improve performance
- Generative production also involves processing a lot of data, but it uses that data as a learning model to then mimic and produce new data.
Why are we talking about AI now?
AI tools have been around for a while, but have mostly gone the machine learning route and have also mostly focused on an enterprise-level customer base.
|💡 Essentially, this makes us part of the world’s largest beta test.
Which underlying marketing needs are AI tools trying to meet?
Few—if any—email marketers ever have enough time. Many of these tools seem aimed at trying to save time through reducing effort.
Email is complex. The phrase, “Just send an email” is a joke in the email community because of all the steps required before we hit send. AI services hold potential for reducing the complexity.
Similarly, AI tools—for better or for for worse—will allow marketers to do more with less because of the previous points, possibly decreasing the cost of sending.
Where do AI tools not meet the hype?
Generative AI tools have already demonstrated a distinct tendency to reinforce racial and gender biases and stereotypes, such as ChatGPT suggesting higher security risk scores to citizens of Syria or Afghanistan when asked how to improve air travel safety, and StableDiffusion only showing white men when asked to produce an image of a CEO.
- Frequent factual errors have been noted by users of ChatGPT, because the data it has been trained upon has not been fact-checked.
- Copy produced by generative AI can come across as generic, lacking a distinct voice.
- Image generators are also able to replicate copyrighted images, which is clearly problematic.
What are the questions we should ask ourselves about AI services?
- Will AI services take my job? Automation and industry have been having all the variations of this conversation over the centuries, and it’s not a question with a neat, binary answer like, “yes, they will take your job,” or, “no, they won’t.” Jobs will change as a result of these services existing, but they are unlikely to replace any jobs outright. What is more likely is that the use of these will allow companies squeeze more productivity out of smaller head counts, for better or for worse.
- Is the proliferation of AI marketing tools inevitable? Not universally, and we—collectively as a community—get a say in which tools become normalized and which are relegated. We’ve seen a number of innovations in our industry (*cough* AMP for email *cough*) that seemed like they would transform the landscape and have been niche use cases instead.
- Which types of these AI tools seem the most helpful? Overall, AI tools that can help iterate, aggregate, or automate seem like the best fits for our industry. Those help minimize the tedious and time-consuming aspects of the work, and mostly tend to fall on the machine learning side of the AI spectrum, as well as services like Subjectgeni.us that are generative but work within distinctly limited parameters.
- Which types should we be skeptical of for the time being? Generative AI services seem mired in potential legal grey areas, produce factual errors, or outright reinforce racism and sexism. In fact, there’s now an open letter signed by pre-eminent AI researchers, AI developers, and tech leaders asking for a pause on further development until adequate regulations can be implemented.
Logan’s final thoughts
In summary, your humble recapper would say that there’s plenty to be enthusiastic about when it comes to AI tools, specifically those that could help us automate the less-stimulating and more effortful parts of email marketing.
Those types of services could free us up to focus on the creative and strategic work that does enliven us and benefits more from our direct involvement.
While the much-hyped generative AI services are worth keeping an eye on, their current lack of regulation and evident beta-status means that mistakes of many sorts are not just possible, but probable.
Above all, let’s be mindful of what we say yes to using—we’ve worked very hard to have the complexity and effort of our labor as email marketers, designers, and developers valued more highly. Let’s keep it that way.
Logan Sandrock Baird is the Senior Community Evangelist at Litmus