What is generative AI?
Let’s start by building a definition of generative AI. Generative AI (GenAI) refers to a category of artificial intelligence that leverages advanced Machine Learning techniques, particularly Deep Learning models, to create new content by learning patterns and structures from existing data. Today, the most advanced generative AI solutions—particularly Large Language Models (LLMs) and diffusion models—are built entirely on Deep Learning, with minimal reliance on traditional Machine Learning approaches. Unlike traditional AI systems designed for specific tasks, such as classification or prediction, GenAI is capable of generating novel content, often mimicking human creativity while producing outputs that are distinct yet similar to its training data.
Now that we know the meaning of generative AI, let’s take a look at how it works on a more technical level. GenAIs use Deep Learning architectures to model complex patterns in data in order to produce new content.
During training, these models analyze large datasets to identify relationships between elements like words, colors, or shapes. These relationships are encoded into structured mathematical representations. Depending on the model type, they can take several forms, such as embeddings in the case of language models, or latent spaces in models for images, audio, and other high-dimensional data.
Techniques like self-attention (used in transformer models) allow AI to focus on the most relevant parts of input data, improving coherence and contextual accuracy. Other architectures, such as GANs or VAEs, use different mechanisms to achieve similar results.
How this works varies depending on the output. For example, in text generation, the model predicts the next word based on preceding words. This is a little different in image generation, where diffusion models create images by progressively removing noise from a random starting state, gradually forming coherent visuals.
This process isn’t perfect and requires constant improvement. Deep Learning specialists employ a variety of methods to improve the outcome of the models. This might involve improving the quality of the dataset, upgrading model architecture, fine-tuning parameters, or a variety of other approaches.
Where did generative AI come from?
One of the mathematical foundations of artificial intelligence can be found in Markov Chains, first introduced by Russian scholar Andrey Markov in 1906. This paper demonstrated how probabilities could describe sequences of events. While early AI in the 1950s and 1960s was primarily based on symbolic reasoning and rule-based systems, probabilistic models like Markov Chains influenced later developments.
Following this, a series of innovations led to breakthroughs in neural networks, starting with the Perceptron in 1958 and the resurgence of Deep Learning in the 1980s with backpropagation. The introduction of Graphics Processing Units (GPUs) in 1999 further accelerated AI development.
In 2014, Ian Goodfellow introduced Generative Adversarial Networks (GANs), which pioneered adversarial training, allowing models to generate increasingly realistic images, audio, and video. While GANs were a significant milestone, later advances in transformer and diffusion models proved to be more impactful for today’s generative AI.
In 2017, Google researchers introduced transformers, a new neural network architecture that revolutionized how AI processes sequential data. Transformers use self-attention mechanisms to analyze relationships between elements in a sequence, enabling models to understand context more effectively than previous architectures. This is achieved by converting text into tokens, which the model processes mathematically to identify patterns and meaning. The introduction of transformers paved the way for today’s generative AI, forming the foundation of Large Language Models (LLMs) like GPT-4.
In 2020, the first advanced generative AI models leveraging transformers and Deep Learning were released. One of the most significant breakthroughs was GPT-3, which set new standards for natural language generation. Over the next few years, multimodal generative AI models emerged, including DALL·E (2021) for text-to-image generation, Stable Diffusion (2022) as an open-source alternative, and Google Bard (now Gemini) in 2023 for conversational AI. Since then, these models have become increasingly sophisticated, driving advancements in foundational AI systems that power a wide range of applications.
Foundational and Task Driven Models
Generative AI models have a number of different classifications. At the highest level, there are two types of generative AI: Foundational Models and Task-Driven Models. Foundational Models are large, general-purpose AI systems that can serve as a base for a wide variety of tasks. This includes some of the most well-known generative AI models, such as GPT-4, which powers ChatGPT. Task-Driven Models are typically narrower in scope and designed to excel at specific tasks or domains. Some are fine-tuned versions of Foundational Models, while others are built from the ground up for specialized applications.
Within these classifications, generative AI is built on different AI approaches, including the following:
Symbolic AI: Uses explicit rules, logic, and symbols to represent knowledge and perform reasoning.
Machine Learning: Enables systems to learn from data and improve performance without being explicitly programmed.
Supervised Learning: Trains models on labeled data to predict outcomes for new, unseen inputs.
Unsupervised Learning: Identifies patterns and structures in unlabeled data without predefined outcomes.
Reinforcement Learning: Teaches agents optimal behaviors through trial and error by receiving rewards or penalties.
While many of the most famous generative AI models are Foundational Models, this isn’t always the case. For example, OpenAI’s Jukebox is a Task-Driven Model designed specifically for music generation rather than general-purpose applications.
Beyond their intended purpose, the key differences between Foundational and Task-Driven Models include training data, architecture, training methodology, and scale. Typically, Foundational Models are trained on diverse datasets, using unsupervised or self-supervised learning, often followed by fine-tuning with supervised techniques. This is a computationally intensive process that can involve billions of parameters. This upfront investment in training makes a foundational model highly versatile, enabling it to be applied across a wide range of use cases.
Examples of Foundational Models include:
GPT-4: A large language model that generates human-like text for diverse tasks, such as writing, coding, and summarization.
DALL·E: An image-generation model that creates unique visuals from textual descriptions, covering a wide range of styles and concepts.
Stable Diffusion: An open-source model that generates high-quality images based on prompts, enabling customization and flexibility in artistic creation.
Claude 2: A conversational AI model developed by Anthropic, designed to generate human-like text and engage in dialogue.
These models are valuable because they allow generative AI to be applied across multiple domains, from text and image generation to automation and personalized content creation.
Generative AI use cases
Generative AI has a wide variety of use cases, both serious and fun. For example, if you’re stuck on what to say or want to explore new ideas, ChatGPT can serve as a brainstorming assistant, offering diverse perspectives and helping you refine your ideas. One thing to note about generative AI tools is that they do not possess human-like creativity but can generate novel outputs by recombining and transforming existing patterns in complex ways. These tools are designed to augment human creativity, not replace it.
GenAI has a broad range of applications across various industries:
Understanding user intent
A key challenge for marketers is interpreting what a user actually wants to do. With third-party tracking cookies being phased out, marketers face increasing challenges in understanding user behavior and reaching the right audience. Without tracking cookies, cookieless marketing on the open web typically relies on context marketing. This requires processing billions of articles to identify the right ones to advertise on.
RTB House’s history of using Deep Learning models laid the foundation for our most recent advancement in GenAI called IntentGPT. This will help marketers understand what their users want without needing to rely on outdated privacy-invasive tracking methods, like third-party tracking cookies.
Generating copy
Generative AI for text can help to create high-quality content at scale. Tools like GPT-4 and Jasper AI can generate blog posts, ad copy, emails, and even social media posts. With the right prompting, generative AI can assist users in navigating their daily lives, especially if they sometimes struggle with finding the right words.
There are also some GenAI tools designed to help with very specific tasks. For example, Persado analyzes emotional triggers to create persuasive ad copy.
Creating and testing visual content
Visual content is a key component of modern life. Generative AI like DALL·E, Canva Magic Design, and Stable Diffusion enables users to produce unique visuals quickly and affordably. This helps people with limited resources get their projects off the ground by allowing them to generate visuals based on simple descriptions without needing to hire an artist at the start. For example, DALL·E can generate an image of a vibrant beach scene or book cover in seconds.
In a corporate context, generative AI tools also simplify A/B testing for visual creatives. By generating variations of designs or ads, marketers can quickly test which resonates most with audiences. Tools like Runway ML allow teams to edit or enhance visuals quickly, making real-time adjustments to campaigns.
Runway ML, Pictory, and Synthesia enable the creation of AI-generated video content, which is increasingly used in marketing, entertainment, and training materials. While video generation is more complex than text or image synthesis, advancements continue to make AI-generated videos more seamless and realistic.
Although this kind of GenAI is impressive, it should be noted that most content for professional purposes still requires post-processing, and human artists remain an essential part of the creative process.
Creating audio for videos
An important use case is the ability to quickly create audio for ads without needing to rely on a voiceover artist. Tools like Descript’s Overdub make it possible for marketers to create synthetic voiceovers for explainer videos or audio-only ads. It can also be used in educational videos, for audiobooks, and a variety of other purposes.
Thanks to these tools, marketers can rapidly localize an ad for different regions by making multiple recordings in different accents, tones, or even languages. This significantly reduces the complexity of international marketing campaigns and makes it easier for smaller brands with fewer resources to reach out to new markets.
What are the benefits of generative AI?
The key benefit of generative AI is improved efficiency. The ability to assist with learning new skills, generate high-quality content, and analyze data can all be hugely helpful. This helps people not only improve their performance, but do so at a higher scale than ever before. This is particularly useful for smaller businesses, new content creators, or someone moving into a new field or hobby.
From a marketing perspective, clever use of Gen AI will help brands communicate in a more personable way. Content localized to the tastes or language styles of a particular region will help them feel more connected to the product and ultimately improve the ad experience. Generative AI can enhance privacy-friendly marketing by enabling contextual targeting, reducing reliance on invasive tracking methods.
What are the limitations of generative AI?
Generative AI isn’t without its challenges. Its impact on certain industries, such as voice acting and writing, has been widely discussed, raising concerns about job displacement and ethical implications. There are also questions surrounding the legality of how the data used to train GPT models was gathered, particularly issues related to copyright and data privacy regulations. However, these aren’t the only challenges posed by generative AI, and the technology itself is limited by a number of factors.
Computing infrastructure
Generative AIs are enormously complicated and require vast amounts of computational power to function. This means that they are expensive to create and maintain. For context, OpenAI was predicted to spend $5 B in 2024. This might be set to change with the rise of new, more efficient GenAI models like DeepSeek, which the company claims costs just $0.14 per million tokens, vs. ChatGPT’s $7.50. However, even with these developments, any serious AI project is still going to require significant computational costs.
Lack of high-quality data
Generative models require high-quality, unbiased data to be effective, and this isn’t always in easy supply. This means that niche use cases may be more challenging to develop because the data to train the AI simply isn’t there. While synthetic data—data produced by another generative AI—can be used to plug the gap, it isn’t a replacement for human-generated training data.
This problem is compounded by the issue of copyright. Organizations have become acutely aware of the value of their data and are making it increasingly difficult for generative AI producers to access it. Additionally, regulations such as GDPR and DMA/DSA create other potential legal challenges for the creators of generative AI tools.
Too much AI content
While AI content is often humanlike, it tends to have its own specific language. This can lead to the homogenization of content, making it easier for consumers to dismiss it as general noise. Additionally, many customers seem to dislike AI content, particularly when a human pretends that they wrote the content. This seems to be because the user feels that it reduces brand authenticity, especially when an AI is used to write about emotional content.
Overreliance on AI tools
Another potential challenge is an overreliance on generative AI tools like e.g., ChatGPT. This can be clearly seen among young students, who have been increasingly using ChatGPT to answer homework questions. One study found that students who used ChatGPT to answer questions performed 17% worse on a test on their subject compared to those who did not have access.
The key is to be careful about how you use AI. Rather than using it to quickly solve problems for you, view it as a partner that helps you come to the correct conclusion yourself.
The future of generative AI
Generative AI will continue advancing, but training frontier models is becoming increasingly expensive. As a result, future improvements may focus on efficiency and specialization rather than sheer model size. We can expect smoother video generation, faster response times for chatbots, and better factual accuracy and reasoning (reducing hallucinations). There will also be more adaptations of models to specific use cases, as we are doing with IntentGPT.
For example, in marketing, we expect to see more marketing teams take advantage of Custom GPTs. These are more tailored versions of standard GPT models, like GPT-4, with behavioral parameters tailored to a specific use case. For example, a company could feed a custom GPT with its brand guidelines and high-quality content examples. It’s a little bit like having a GPT with all your custom prompts permanently applied.
With time, this new custom GPT would then be able to take the best aspects of high-performing content and produce new, unique content in the brand’s voice. This reduces the amount of editing needed by the marketing team, and helps to mitigate the problem of too much AI content.
Learn how your company can implement generative AI today
The RTB House team has always put artificial intelligence at the center of its operations. We already use Deep Learning in all of our marketing services. We’re also actively developing IntentGPT, a GPT model designed to help understand how users interact with open web content and help companies reach them with effective personalized content.
If you’d like to learn more about how your company can run personalized ads powered by Deep Learning with RTB House, contact us today.
