Generative AI and big language models are set to change the marketing industry as we know it.
To stay competitive, you’ll need to understand the technology and how it will impact our marketing efforts, said Christopher Penn, Chief Data Scientist at TrustInsights.ai, speaking at The MarTech Conference.
Learn ways to scale the use of large language models, the value of rapid engineering, and how marketers can prepare for the future.
The premise behind the great linguistic models
Since its launch, ChatGPT has been a trending topic in most industries. You can’t go online without seeing everyone’s opinion. However, not many people understand the technology behind it, Penn said.
ChatGPT is an AI chatbot based on OpenAI’s GPT-3.5 and GPT-4 Large Language Models (LLM).
LLMs are built on a premise from 1957 by the English linguist John Rupert Firth:
“You’ll know a word of the company he keeps.”
This means that the meaning of a word can be understood from the words that usually appear next to it. Simply put, words are defined not only by their dictionary definition, but also by the context in which they are used.
This premise is key to understanding natural language processing.
For example, look at the following sentences:
“I’m making tea.” “I’m spilling the tea.”
The former refers to a hot drink, while the latter is slang for gossip. “Tea” in these cases has very different meanings.
Word order also matters.
“I’m making tea.” “The tea I’m making.”
The sentences above have different focus topics, even though they use the same verb, “to brew.”
How the big language models work
Below is a diagram of the transformer system, the architectural model on which large language models are built.
There are two important features here inlays i positional coding.
Simply put, a transformer takes one input and turns it (that is, “transforms”) it into something else.
LLMs can be used to create, but they are better at turning one thing into another.
OpenAI and other software companies start by ingesting a huge corpus of data, including millions of documents, academic papers, news articles, product reviews, forum comments and much more.
Think how often the phrase “I’m making the tea” might appear in all these ingested texts.
Amazon product reviews and Reddit comments above are some examples.
Notice the “company” this sentence retains, meaning all the words that appear near “I’m making the tea.”
“Taste”, “smell”, “coffee”, “aroma” and more give context to these LLMs.
Machines can’t read. So to process all this text, they use inlaysthe first step in transformer architecture.
Embedding allows models to assign a numerical value to each word, and this numerical value appears repeatedly in the text corpus.
Word position is also important for these models.
In the above example, the numeric values are still the same, but they are in a different sequence. This is positional encoding.
In simple terms, large language models work like this:
Machines take text data. Assigns numeric values to all words. Look at the statistical frequencies and distributions between the different words. Try to figure out what the next word in the sequence will be.
All of this requires significant computing power, time, and resources.
Get the daily search newsletter marketers trust.
Rapid engineering: a skill that must be learned
The more context and instruction we provide LLMs, the more likely they are to achieve better results. This is the value of rapid engineering.
Penn thinks of the directions as run-ins for what the machines will produce. Machines will pick up the words in our input and latch onto them for context as they develop the output.
For example, when writing ChatGPT requests, you’ll notice that detailed instructions tend to return more satisfying responses.
In a way, prompts are like creative briefs for writers. If you want your project to be done right, you won’t give one-line instructions to your writer.
Instead, you’ll send a decent-sized summary that covers everything you want them to write and how you want them written.
Expanding the use of LLM
When you think of AI chatbots, you may immediately think of a web interface where users can enter requests and then wait for the tool to respond. This is what everyone is used to seeing.
“This is not the end game for these tools by any means. This is the playground. This is where humans get to play with the tool,” Penn said. “That’s not how companies are going to bring this to market.”
Think of speed typing as programming. You are a developer who writes instructions for a computer to do something.
Once you’ve tailored your requests for specific use cases, you can leverage the APIs and have real developers wrap those prompts in additional code so you can programmatically send and receive data at scale.
This is how LLMs will scale and change companies for the better.
As these tools are being implemented everywhere, it’s critical to remember that everyone is a developer.
This technology will be in Microsoft Office – Word, Excel and PowerPoint – and many other tools and services we use every day.
“Because you’re programming in natural language, it’s not necessarily traditional programmers who are going to have the best ideas,” Penn added.
Because LLMs are driven by writing, marketing, or PR professionals, not programmers, they can develop innovative ways to use the tools.
How LLMs will affect search marketing and what you can do about it
We’re starting to see the impact of big language models on marketing, specifically search.
In February, Microsoft introduced the new Bing, powered by ChatGPT. Users can converse with the search engine and get direct answers to their queries without clicking on any links.
“You should expect these tools to take advantage of your unbranded search because they answer questions in a way that doesn’t require clicks,” Penn said.
“We’ve faced this before as SEO professionals, with featured snippets and click-free search results…but it’s only going to get worse.”
He recommends going to Bing Webmaster Tools or Google Search Console and looking at the percentage of traffic your site receives from unbranded informational searches, as this is the area of greatest SEO risk.
Build your brand
“If branding isn’t one of your top strategic priorities for 2023 and beyond, it needs to be,” Penn stressed.
You need to build your brand and get people asking for your name in search.
When users ask for ideas or recommendations on a topic, LLMs will likely direct them to the synthesized information, not you.
But if people specifically ask for your brand by name, they’ll still get where they want to go.
Make your brand’s online presence as strong as possible.
Use a publishing platform that is “immune” to AI
Penn also emphasized the importance of using a platform where you have direct, unmediated access to your audience.
Channels like email or SMS (even direct mail) allow you to reach out to customers directly and make sure you reach them without being mediated by AI.
Organic search and social media are already heavily mediated by AI. Therefore, the likelihood of reliably reaching even a fraction of your audience is slim.
Even the biggest brands can only get enough views if they spend on paid campaigns.
Services like Slack, Telegram, and Discord allow you to meet like-minded people and develop meaningful connections.
When you provide value to your users, you can reliably reach them, earn their loyalty, and build brand equity.
See: The Marketing Singularity: Great Linguistic Models and the End of Marketing as You Know It
Penn shared more about the impact of LLMs on marketing jobs at The MarTech Conference. Watch his full presentation here:
[ad_2]
Source link
