Under a rock.

That's where you've been living this past week if you haven't heard about ChatGPT by now. ChatGPT is OpenAI's latest release of large-language models. It's impressive, and it's caused quite a stir. Some people say it will kill Google, which is quite the claim, but there are many reason why you should pay attention to ChatGPT.

There’s little doubt that AI’s impact on marketing will be profound, but is ChatGPT the game changing tool that many people claim it to be?

If ChatGPT is so clever, I thought I'd ask ChatGPT to write an article about large language models and get it to speculate on the impact on its very existence. In the prompt, I wrote:

Write an article for my technology reporting website about the history of AI language models; talk about GPT3 and other models, such as AI21's Jurassic model, and explain why Google's search dominance could be under threat due to models like ChatGPT.

ChatGPT replied:

As AI technology continues to advance, language processing has become a crucial area of focus for many researchers and developers. In recent years, we have seen the emergence of powerful AI language models that are capable of processing and understanding human language, allowing them to perform a wide range of tasks.

In this article, we will explore the history of AI language models, from their early beginnings to the state-of-the-art models of today. We will also look at some of the most notable examples of AI language models, such as GPT-3 and AI21's Jurassic model, and discuss how these models are being used in various industries. Finally, we will touch on the potential impact of AI language models on the search engine market, and why Google's dominance in this area could be under threat.

The history of AI language models

The idea of using computers to process and understand human language has been around for decades, but it was not until recently that we have seen significant progress in this area. In the early days of AI research, language processing was seen as a difficult task due to the complexity and nuances of human language. However, advances in machine learning and neural networks have enabled researchers to develop AI models that are capable of processing language in a more human-like way.

One of the key milestones in the history of AI language models was the development of GPT-2, a language model developed by OpenAI in 2019. GPT-2, or Generative Pretrained Transformer 2, was trained on a massive amount of text data, allowing it to generate human-like text that was often difficult to distinguish from text written by a person. This marked a significant step forward in the field of AI language processing, and paved the way for further advancements.

Notable AI language models

Since the development of GPT-2, there have been many other notable AI language models that have been developed. One of the most impressive examples is GPT-3, the successor to GPT-2, which was released in 2020. GPT-3 is even larger and more powerful than its predecessor, and has been used to perform a wide range of tasks, from translation and summarization to text generation and question answering.

Another notable AI language model is AI21's Jurassic model, which is specifically designed for long-form text generation. Jurassic is capable of generating entire articles on a given topic, complete with accurate grammar and structure. This makes it a useful tool for content creation, and it has been used by companies to generate product descriptions and other written materials.

Use cases for AI language models

AI language models have a wide range of potential use cases, and they are being used in many different industries. For example, GPT-3 has been used to create chatbots that can hold natural conversations with users, and it has also been used in the development of virtual assistants like Siri and Alexa. AI language models are also being used in the field of education, where they can help students learn new languages by providing personalized feedback and corrections.

One of the most interesting use cases for AI language models is in the field of programming. Using natural language prompts, developers can use AI language models to generate code that is tailored to their specific needs. This can save time and reduce the need for manual coding, making it a valuable tool for developers.

The impact on the search engine market

As AI language models continue to improve, they have the potential to disrupt many industries, including the search engine market. With their ability to understand and generate human language, AI language models could potentially provide more accurate and relevant search results than traditional search engines. This could pose a threat to the dominance of search giants like Google.

What does this sample article show us about the capabilities of ChatGPT? Did it do an excellent job of responding to the original prompt? Is the reply wholly accurate?

This sample article shows us that ChatGPT can produce detailed and seemingly accurate prompt responses. It responded well to the given prompt, providing a brief overview of the history of AI language models and discussing notable examples and their potential applications.

The article was largely accurate, although there may be some slight discrepancies. For example, the LLM model Jurassic was not, as far as I am aware, developed explicitly for long-form text generation but is in fact, a versatile, general-purpose foundation model capable of performing many tasks.

We also see that the model can speculate on a scenario. It can invent scenarios, based on nothing more than a suggestion, as demonstrated by its output regarding the impact on the search engine market.

Ultimately, that’s the key thing to remember. AI can speculate. It can make stuff up.

Or, as Cassie Kozyrkov, Chief Data Scientist at Google, said:

There’s something very important you need to know:

ChatGPT is a bullshitter.

Martin Broadhurst
Post by Martin Broadhurst
December 8, 2022
Martin Broadhurst is a sales and marketing technology consultant with specialising in HubSpot and Marketing AI technology.