Everyone is talking about artificial intelligence. It makes sense — suddenly there are free (or cheap) tools readily available that let you create a variety of AI-generated content, including text and images, in seemingly unlimited styles in seconds.
Exciting of course.
But stop for a moment and ask yourself a few questions.
- Do I Really Know What AI Is?
- Do you know how long it’s been around?
- Do you know the difference between AI and machine learning?
- And do you know what deep learning is all about?
If you answered affirmatively to all these questions, this article may not be for you. If you’re hesitant about some of them, read on.
The AI Revolution Begins… Now?
Let’s start by filling in the background.
Is AI new?
no. At least conceptually, AI dates back to his 1950s (more on that later). In the 1960s and 1970s, as computers became faster, cheaper, and more widely available, it began to flourish as a practical pursuit.
Is AI in Marketing Anything New?
no. It’s worth remembering that AI has long had so many applications in marketing beyond content creation. Content and product recommendations have been powered by AI for years. Predictive analytics — Predict user behavior based on large datasets of past behavior and predict the next best thing (show relevant white his paper, show red baseball cap, send email) Used to — AI-long drive.
Well-known vendors have been incorporating AI into their solutions for almost a decade. Adobe Sensei and Salesforce Einstein date him back to 2016. Oracle’s involvement with AI goes back at least. I never gave it a cute name. Another veteran deployer of AI is Pega, who first uses it to predict the next best action in his business process management product and then in his CRM platform.
So… is generative AI something new?
Generative AI. Conversational AI. AI writing tool. All the phrases in that moment all have overlapping meanings. Generative AI produces text (or images, or even videos). Conversational AI interacts with a human interlocutor to generate text (think AI-powered chatbots). AI writing tools are intended to create customized text on demand. All of these solutions use “prompts” in some way. That is, wait to be asked or set a task.
Is this all new? no. What is new is its wide availability. Natural Language Processing (NLP) and Natural Language Generation (NLG) have been around for years. The former means his AI interpretation of the text. The latter is AI-powered text creation. Back in 2015, based on my own reports, his AI-powered NLG was producing written reports for doctors and industrial operations. Additionally, he produced weather forecasts for the United Kingdom Meteorological Service, the National Meteorological Service.
Data input, text output. It’s not as widely available as something like ChatGPT.
video too. By 2017, at least, we will be using AI to create video content that is not just personalized, but individualized. It’s generated when the user clicks play, so it’s fast enough that it looks like it’s streaming from your existing video library. Again, not widely available, but rather an expensive enterprise product.
Dig deeper: Chat GPT: Marketer’s Guide
What is AI: The Simple Version
Let’s start with the basics.
start with the algorithm
Algorithms can be defined specifically as a set of rules that computers follow in computations and other problem-solving or task-completion operations. Is “algorithm” from Greek? No, it actually comes from part of the 9 name (al-Khwārizmī).th Arab mathematician of the century. But it doesn’t matter.
The point is that using algorithms for calculations and tasks is not the same as using AI, not repetition. Algorithms are easy to create. Let’s look at a simple example. Suppose you run an online bookstore and want to offer recommended products. You can create 100 rules (algorithms) and train your website to follow them. “If she searches for Jane Austen, also show her Emily Brontë.” “If he’s looking for Agatha Christie, show him some other detective story.”
Of course, you’ll have to tag your detective volume appropriately, but so far it’s been easy. On the one hand, these are good rules. On the one hand, they are not “intelligent” rules. That’s because they are fixed unless I go back and change them. If someone looking for a WW1 book consistently ignores a WW2 book, the rules will neither learn nor adapt. They keep doing what they are told.
With Amazon’s resources, make your rules intelligent. This means it can be modified and improved according to user behavior. If I had Amazon’s market share, there would be a ton of user behaviors that rules could learn.
If an algorithm can learn on its own, with or without human supervision, then we have AI.
wait a minute. Isn’t it just machine learning?
AI vs machine learning
For purists, AI and machine learning are not inherently the same thing. But as big as it is, these terms are used interchangeably, so there’s no turning back. Instead, the term “AI in general” is used when people want to talk about pure AI, AI in the true sense of the word.
Let’s go back to 1950 (I warned you to do so). Alan Turing was a brilliant computer scientist. He helped the Allies defeat the Nazis through code-breaking espionage. His reward was the abhorrent treatment by British society for his (still illegal) homosexuality, and more than 50 years after his death he brought an official apology from Prime Minister Gordon Brown. It was handled well. I am very proud of my freedom thanks to Alan’s work. You deserved better. ”
But what about AI? In 1950, Turing published his groundbreaking paper, Machines and Intelligence in Computing. He published it in the philosophical magazine “Mind”, not in a scientific journal. At the heart of this paper is a kind of thought experiment he called “The Imitation Game”. It is now widely known as the “Turing test”. Simply put, it proposes a standard for machine (or artificial) intelligence. Intelligence can be attributed to a machine if a human interlocutor cannot distinguish between a machine’s response to his question and another human’s response.
Of course, Turing’s proposal has a great many objections (and his tests are not smartly designed). But this started the quest to replicate human intelligence, or at least create something equivalent. IBM Watson can be seen as a continual pursuit of that goal (although there are many less ambitious and more profitable uses of his).
No one expects product recommendation machines like Amazon or content creation engines like ChatGPT to be as intelligent as humans. For one thing, they can’t know or care if what they’re doing is right or wrong. They act on data and predictive statistics.
In fact, all AI discussed here is actually machine learning. But I’m not going to stop people from calling it AI. When it comes to pursuing human-level or “general-purpose AI,” there are good reasons to believe that it is not just around the corner. See, for example, Erik J. Larson’s The Myth of Artificial Intelligence: Why Computers Can’t Think Like Us.
What about “deep learning”?
You may also come across “deep learning” as an AI-related term. Is it different from machine learning? Yes, it is; this is a big step beyond machine learning, and its significance is that it greatly improves AI’s ability to detect patterns, making images (and videos) as well as it processes numbers and words. ) can now be processed. This gets complicated. here’s a short version.
Deep learning is based on neural networks. A neural network is a layer of artificial neurons (bits of mathematics) that are activated by an input, which communicate with each other and produce an output. This is called “forward propagation”. As with traditional machine learning, the node looks at how accurate the output is and adjusts its operations accordingly. This is called “backpropagation” and trains the neurons.
However, between the input and output layers there is also a multiplication of what is known as a “hidden layer”. Think of these layers literally stacked on top of each other. That’s why this kind of machine learning is called “deep”.
A stack of network layers turned out to be much better at recognizing patterns in input data. Deep learning helps with pattern recognition. This is because each layer of neurons decomposes complex patterns into simpler patterns (the backpropagation training process is also ongoing).
Are there AI vendors in the martech space?
it depends on what you mean.
AI-powered vendors
There are an estimated 11,000+ vendors in the martech space. Many of them are probably mostly AI powered (or you could very well claim they are). But they aren’t using AI for its own sake. they are doing something with it.
- To make commercial recommendations.
- To write the subject of an email.
- Recommend next best actions to marketers or sales reps.
- To power our chatbots.
- Writing ad copy.
- To generate content for large-scale multivariate testing.
The list is endless.
The point I want to emphasize is that AI is a bit like salt. Salt is added to food to improve its taste. At least most of us prefer to use proper salt in our diets. But no one says, “I have salt for dinner” or “It feels like a snack.” I’d like some salt. ”
We add salt to our food. Incorporate AI into your marketing technology. Salt and AI are probably not often used alone, except for research purposes.
Yes, there are countless martech vendors using AI. But are there any martech vendors selling AI as a separate product?
Vendors selling AI
The answer is that there are very few in the martech space. AI as a product really means AI software designed by engineers that can be used and embedded in the context of other solutions. It’s easy to find engineering vendors that sell AI software, but in most cases they sell to IT organizations rather than marketing organizations, and they have a very broad back ground rather than enabling marketing or marketing. Sold for office use. sale.
With one or two exceptions, they are clearly targeted at marketers. However, it’s not enough to create a populous category in the world of marketing technology.
we scratched the surface
That’s all this article is about. With a rich history behind it, we can scratch the surface of a highly complex topic with an unpredictable future. Of course, there are ethical issues to deal with, such as the nearly inevitable case of machine learning models being trained on skewed datasets, and the equally inevitable case of human content theft by generative AI.
But hopefully this is enough for you to chew on for now.
Get Martech! every day. free. to your inbox.
