Everything changed when OpenAI’s ChatGPT publicly debuted in November 2022. Within just three months the platform broke the record for the most rapidly adopted software ever with an incredible 100 million plus users in that short space of time.
Those initial months witnessed a real frenetic scramble to harness ChatGPT’s capabilities across a diverse range of applications. From individuals striving to master AI prompting to the automation of tasks like crafting cover letters, resumes, and even academic papers, the AI tool’s sudden versatility was undeniable.
Notably, OpenAI was challenged by Bill Gates to test its AI’s aptitude with the Advanced Placement Biology exam. The results, with ChatGPT answering 59 out of 60 questions correctly, and earning an A or A+ grade, also saw the AI tool receiving praise for its articulate responses to open-ended inquiries.
Then, fuelled by substantial investment from Microsoft, ChatGPT was swiftly integrated into Bing, instantly reshaping the search engine experience.
Google’s Response: Enter the Bard
While OpenAI’s advancements were widely celebrated, they also highlighted the challenges faced by Google with their long history of more subtle AI integration into search, exemplified by technologies like BERT and MUM. The foundation of these advancements lies in the strides made in Natural Language Processing (NLP) and Google’s own ground-breaking LaMDA technology in 2021.
The emergence of ChatGPT triggered a “code red” strategic response at Google, with founders Sergey Brin and Larry Page actively involved in shaping the company’s AI-driven chat strategy.
Gemini, Google’s most capable and general model, was unveiled in December and began rolling out in Bard shortly after.
The Impact of LLMs on Digital Marketing Strategies
While the technological feats of these LLMs are impressive, my focus has always remained on delivering effective digital marketing solutions for local businesses.
I initially explored the potential of LLMs by generating content for a local digital marketing agency’s blog, aiming to streamline the creation of a backlog of articles…
Leveraging LLMs for Content Generation
I tasked AI with creating a series of articles on diverse topics relevant to the digital agency’s audience.
The Upsides
- Speed: The efficiency of the content generation was impressive, with articles produced in a fraction of the time required for traditional long-form writing.
- Coherence: The AI generated text was generally coherent and logically structured.
The Downsides
- Repetitive Structure: The output tended to follow a predictable format, often consisting of introductory paragraphs, bullet points, and a concluding summary. The language was repetitive too.
- Limited Depth: AI generated articles were typically concise, generally output at 300 words, the absolute minimum for meaningful reading and often labelled as “thin content”.
- Lack of Originality: The generated content lacked the unique style and perspective that human writers bring.
- Absence of Expertise: The AI could certainly not incorporate personal experiences or industry insights into its writing. That’s where being human is irreplaceable.
- No Citations: The generated text lacked references to supporting sources. This is changing however, and differs between AI engines and models.
- Outdated Information: The data that the models are trained on has a cut off point. ChatGPT’s initial training data went up to October 2022 IIRC.
- Inconsistent Knowledge: The models can struggle with niche or very specific queries.
The latest models are a vast improvement and many of these initial issues are being changed on a regular basis as users demand more from these new AI tools.
Our experimentation with Google’s Gemini yielded similar results.
Utilising LLMs for Keyword Analysis
Beyond content generation, I explored the application of LLMs in keyword research.
My current workflow involves a combination of tools like Google Search Console, Bing Webmaster Tools, and Google Ads Keyword Planner, supplemented by my extensive experience in SEO, PPC, and content strategy.
Now with Bard and ChatGPT these LLMs enable us to:
- Generate lists of relevant keywords for specific locations and services.
- Retrieve keyword search volume data.
- Identify long-tail keywords and variations.
- Categorise keywords based on search intent.
- Retrieve keyword counts.
This new streamlined approach has reduced the need to navigate multiple platforms although double-checking output against them is incredibly useful.
LLMs for Crafting Titles and Meta Descriptions
Creating effective title tags and meta descriptions is a critical SEO task. AI can assist in generating creative ideas and expanding vocabulary.
While experienced digital marketers can handle these tasks proficiently, LLMs offer a valuable tool for brainstorming and overcoming creative blocks. I suggest using AI to generate volumes of titles and descriptions but with final oversight and editorial from an experienced SEO professional.
AI-Powered Ad Copy Generation
LLMs can be effectively used to generate alternative ad copy for PPC campaigns.
By providing AI tools with existing ad copy, we can generate variations that may offer fresh perspectives or highlight overlooked opportunities, much in the same way that I suggested for titles and descriptions.
Preliminary Assessment
While LLMs are often referred to as “artificial intelligence,” their capabilities are primarily based on pattern recognition and data analysis. They are powerful language models, but they require careful prompting and potentially further training to achieve optimal results.
LLMs have proven useful in streamlining keyword research and generating ad copy variations.
However, content generation remains a complex area, with ongoing debates about the ethical implications and potential for low-quality content proliferation.
The risk of a resurgence of “content farms,” where websites are flooded with AI-generated content of questionable value, is a concern. This concern has been addressed by Google in the March 2024 core update with a crackdown on
LLMs can be valuable tools in the hands of both novices and seasoned professionals. However, their effectiveness hinges on the user’s ability to leverage them intelligently.
Ultimately, LLMs are tools, and their impact depends on how they are used.
The Human Element in Digital Marketing
While AI offers valuable assistance, I’m committed to providing human-generated content as much as possible for human consumption. This article initially appeared elsewhere but my original attribution has been deleted so I’ve had to have this rewritten by AI and edited by myself. So this is all my work, re-worked by AI, and hand-edited by yours truly once more. If you need a digital marketing expert who uses AI intelligently and in a human way, call me now on 01252 692 765 or leave a message via my contact form.