As digital marketers, we’ve been at the forefront of the deployment of artificial intelligence technology, from generating Google Ads text and social media posts to image generation and even website building and coding. But there are deeper implications to this technology than assisting with blog article writing.
On the road to achieving artificial general intelligence (AGI), researchers began building machine-learning algorithms that can understand and replicate human language. And while the technological breakthroughs enabling machine learning and artificial intelligence have been astounding, the most newsworthy developments in AI have been in large language models — like Chat GPT, Meta’s LLaMA, and Google’s PaLM-2.
In this article, I’m going to explore the impact that large language models have had on the marketing industry, especially in search engines. I’ll also consider some future implications based on the latest AI research, and share my top five AI tools for workflow efficiency.
A Quick History of Search Engines and Artificial Intelligence
When Google first appeared on the scene in 1998, it used a simplified algorithm to measure the number of links a website received from other websites and read the content looking for specific keywords that matched queries.
Since then, the technology behind Google Search has constantly improved. Google still has more than 92% market share of all internet searches, about 100,000 searches per second.
In 2015, Google introduced RankBrain, a machine-learning system that categorizes and works to understand internet content. Machine learning, at its core, is a subdomain of artificial intelligence that deals with computer systems able to extract their own knowledge from a dataset without having that knowledge pre-programmed.
You’ve been using them for years. Pandora, Spotify, Netflix, and Google — all have machine-learning algorithms that predict what you’re looking for.
In 2019, Google pushed a massive update to its search algorithm called BERT, an open-source machine learning framework for natural language processing. This was a huge shift from using keyphrases and backlinks to rank content to understanding the content and how it relates to the intent behind the user’s query.
And in 2023, Microsoft released something they’ve been working on for the last two years: Bing search integrated with Chat-GPT. In what felt like a direct challenge to Google’s supremacy, Bing offered a glimpse into a future where search engines conversed rather than merely retrieved.
This brings us to the present day.
AI Will Change Everything
A profound shift in computing is happening. And it’s happening right now.
According to a study by OpenAI and the University of Pennsylvania, AI is poised to affect nearly 80% of jobs in the United States. Primarily, AI will be used to streamline repetitive tasks, assist creative workflows, and manage large amounts of data. Think: project management, marketing reports, SEO tasks, internal presentations, social media schedules, and paid ads automation.
But that’s not all. The study declared that large language models (LLMs) like GPT-4 are general-purpose technologies, in other words, LLMs have the capability to drastically alter society, just like electricity, the internet, and airplanes.
It’s easy to think of some first-order effects that AI will transform in society, for example: driving cars, trucks, and airplanes. But those will cause 2nd and 3rd order effects like reducing the amount of motor vehicle crashes on the road, which will cause a shortage of organ donors.
But what about the marketing world? As functional content is transformed into a commodity, it may not make sense to hire copywriters to write your website copy anymore. But editorials, opinions, complex and fast-moving topics — these are areas where creatives will still crush the AI-generated content.
It’s easy to be cynical of new tech as it comes out, but the true benefit comes with constant experimentation of new tools that could revolutionize your business, so you’re an early adopter. My philosophy on marketing is constant experimentation, which means testing new tools as they roll out and rapidly deciding whether it improves my workflow or not.
One of the most interesting and impactful shifts in computing is happening in the way we search the internet. Both Google and Microsoft have announced AI-connected search on Google and Bing, introducing a profound shift in the way we search the internet.
Let’s look at Google, for example.
Google’s Generative Search Experience
On May 10th, 2023 during Google I/O, Google dropped a new experiment called generative search experience, which uses its latest AI engine, PaLM-2, to generate snippets summarizing an answer to your query. In seconds, AI analyzes and combines the top search results to give you personal, customized answers to your particular question.
In this example, I asked: “How will AI change Google Search?”
The generative AI looks at the top-ranking Google results and summarizes an answer to your query. It also provides links to the top relevant results suggests contextual follow-on questions.
Let’s look at what it says:
- Reduce the number of steps to answer your query (zero-click results)
- Analyze your behavior and search history to give you better results
- Combine and summarize information across multiple websites
- AI text snippets
There are a ton of implications for search engine optimization, and for the way people will experience your business’ presence on the internet and social media.
But this is just the start.
Google announced lightweight models of PaLM-2 that can be used on a Pixel smartphone without an internet connection. Open-source AI models are also being developed more rapidly and with more highly-tuned features than Google or OpenAI combined.
Imagine: your own instance of AI tuned to your personal data. It will be able to add events to your calendar, give you reminders of things you’re likely to forget, know when you need milk, bread, and eggs — and even replicate your tone of voice across emails, company presentations, and phone calls.
What’s more, it will know when you need to take a break because you didn’t get enough sleep last night or when you’re stressed out based on your biometric markers.
But let’s get back to Google and how we’re going to adapt to the changes from PaLM-2.
AI Search Engine Optimization
With Google’s generative search experience, we’ll need to optimize content for AI snippets instead of featured snippets. Google is rapidly expanding zero-click results.
This means providing answers in their search product rather than having users click through to your website. Google’s business is structured around monetizing search results with ads, so the longer they can keep you on Google, the more impression-share they can sell.
This means instead of users navigating to your website for information, the data will be rewritten by Google to match the specific intent behind the user query. In the future, this will be even more personalized. For example, Google will know you have a wife and three kids, so when looking for hotels at a vacation destination, it will tailor results to your individual needs.
That will be pretty cool individually, but as marketers, how will that impact our organic traffic?
If you’re following current best practices, I doubt you will see a big shift in organic rankings. But traffic? If you’re not in the AI snippet, you might drop a lot of traffic.
But on an entirely different level, AI systems are going to truly begin to understand how users interact with your content and promote content that might not have best-in-class SEO but entices users to go deeper.
Because it’s all about human-to-human interaction.
As it becomes easier to churn out blog posts using AI, the types of content that will succeed are editorials, case studies, tutorials, newsjacking, and video. Most large language models are trained on text from the internet and books, meaning it won’t be able to have a unique opinion or be truly creative — just derivative of all other content out there.
Remember: write for your ideal customer, answer specific queries relevant to your business, and ensure your title tags, headings, and schema are all handled. The AI search algorithms are designed to bring more relevant results to searchers, so it’s our job to create compelling content and tell a story that draws people in and brings them back to interact with our brand.
What Does AI Mean for Financial Services Marketing?
We will need to develop ways to entice the user to click past search results. We should optimize blog posts with creative copy, a clickable image, and a video — and perhaps distribute our content on social media more often and with more budget.
But there’s good news for fintechs. Google’s search generative experience won’t impact Your Money or Your Life content like finance, insurance, and medicine — yet. Google doesn’t want its AI systems to “hallucinate” incorrect finance or medical advice.
This means that financial services marketers will have time to see how AI in search impacts other niches and enable us to develop best practices for optimization before our sites are affected.
But the future is clear. Artificial intelligence is going to permanently change the way search engines work, and the way we optimize our content to be found online.
It’s also going to help us handle larger, more complex clients with ease.
AI Marketing Tech
Some of the coolest developments using LLMs have been for marketers. And it’s not just generating blog articles (this one is written the old-fashioned way, btw). It’s worth exploring new platforms and apps that could impact your ability to handle heavier workflows and streamline existing processes.
Here are a few of my favorite tools:
PageGPT generates high-quality landing pages from a text prompt. Watch the demo — it’s pretty slick.
Stockimg AI generates book covers, stock photos, and more from a text prompt. I’m excited for the day we can generate marketing images with tried-and-true conversion-based copywriting tuned on your own brand voice.
Auto-GPT is an open-source AI model that chains Chat-GPT responses together, asking for reasoning, logical next steps, and processes to achieve a certain goal. It seriously pushes the boundaries of what is possible with LLMs. My favorite UI for Auto-GPT is godmode.space.
I used Auto-GPT to write a Python script that would crawl a directory website and output company names, websites, and contact information to a CSV file. It researched which Python libraries to use, wrote and tested the code, and then even told me how to install Python on my Mac and run the script. All I had to do was pick the right CSS selector from the website code that corresponded to the company name and contact info.
Next-level stuff that gives us a hint of what’s to come.
Playtext.app is an AI reading assistant. It extracts text from a website, then presents it in a distraction-free (no ads) user interface, and generates an AI voice while highlighting the text. It helps you train to read up to 3x or 4x your normal speed.
Jasper.AI is one of the original GPT-3 copywriting tools. They’ve recently introduced cool features like brand voice, target audience, and zoom transcript summarization. I often use Jasper to help with brainstorming content ideas, rewriting sections of text I don’t like, and to help outline video content or blog articles as a starting point.
Google Workspace Duet AI is newly launched in beta. There are a myriad of tools that should help your company’s workflow. There’s the obvious stuff like copywriting and stock image generation, but also some powerful tools coming to Google Sheets, like entering text prompts to create sophisticated outputs like schedules and custom calculations.
There has been a flood of AI-based marketing tools this year because AI is changing rapidly. It took OpenAI two years to design and train GPT-3. When Facebook announced its LLaMA language model in March of 2023, it was quickly leaked to 4Chan and then released as open source.
Within six weeks, open-source developers had trained models to be just as good as Chat GPT. They were also able to create lightweight language models that could run on an Android phone. And they did it 17x faster than OpenAI!
Open source AI plus the ability to access the API for Chat-GPT and GPT-4 will enable developers to build on top of the capabilities of large language models — creating emergent features and capabilities we could only dream of.
I’ll be on the lookout for new AI tech that transcends the ability to write Google Ads or custom CSS for your webpage. When a platform comes out that can take a video and break it into 25 social media posts, complete with copywriting, UTM links, and automatically schedules them based on community engagement — it’s a game-changer for early adopters.
Looking Deeper: Past the Hype
It’s easy to be skeptical.
When the automobile industry started to hit the media cycle, reporter E.P. Ingersoll wrote: “The notion that electric vehicles, or vehicles of any other kind, will be able to compete with railroad trains for long-distance traffic is visionary to the point of lunacy.”
He continued to say that if vehicles were mass adopted, “it will be in a world where natural laws are all turned topsy-turvy, and time and space are no more.”
A lot of articles written about AI are heavy with cynicism. And while most don’t assume that disruption of the space-time continuum is necessary for mass adoption, most seem to miss the point entirely — AI is here to stay, and this is only the beginning.
Mass adoption of AI is going to create a profound shift in the way we interact with computers and the internet.
Right now, it means asking deeper questions about our internal processes. In what ways can our work be streamlined?
- Can we automate SEO tasks like keyword research and metadata analysis?
- Can our writers use AI to assist in their writing process, especially if connected to the internet and SEO tools?
- Can we use AI to analyze our content and suggest improvements based on competitor rankings?
- In what ways can project management be streamlined? What about account management? Paid media automation?
And in the future, it means taking a look at the content we’re putting out and elevating it to the next level. As website copywriting and blog production get increasingly commoditized, it will give room for creators to thrive. Video, editorials, and tastemaking will be able to shine because AI systems are trained on currently existing data. You’ve got to be putting out content that doesn’t exist anywhere else.
Because, ultimately, we’re building human-to-human relationships through the digital world. We do that through authenticity, service, and a willingness to take a risk. And that’s something that an AI isn’t even close to understanding.