AI Scribe - AI SEO Content Creator, Blog Writer, Humanizer & SEO Assistant ChatGPT (OpenAI GPT-4o 128K)

AI Scribe integrates with OpenAI's GPT-3, GPT-3.5 Turbo, and GPT-4 engines to generate content based on your input and preferences. It provides prompts and suggestions to help you create SEO-optimised content efficiently.

AI Scribe itself is free to download and use, but you will need access to OpenAI's API, which may have associated costs depending on your usage. Please check OpenAI's pricing for more details: https://openai.com/pricing

AI Scribe allows you to edit prompts at each stage of the content creation process. You can use your own instructions and pre-defined shortcodes/parameters, such as [Language], [Style], and [Keywords Selected]. Custom default prompts can also be saved within the global settings page. As well as custom prompts, you can now also add custom instructions to be applied alongside all prompts with GPT-3 and GPT-4 models.

Custom instructions in ChatGPT allow you to guide the model's behaviour during a conversation. These instructions can be used to specify rules, guidelines and aspects of the generated text. For AI-Scribe, the custom instructions take advantage of OpenAI's "system" role within the messages you exchange as a "user". This only works for GPT-3 and GPT-4.

AI Scribe provides configurable options for language, style, and tone. By selecting the appropriate settings, you can create content that matches your brand's voice and tone, ensuring a consistent user experience across your website.

Yes, AI Scribe offers multiple usage options, including saving the content as a shortcode. This allows you to use the generated content within widgets, modules, and page builders like Elementor and Divi.

AI Scribe provides evaluation features that assess your content's quality, although these results should be taken with a pinch of salt. Future updates plan to integrate third-party tools for a more accurate and detailed evaluation.

Token issues may occur for very long articles or when using specific AI models. To resolve token issues, consider shortening the content or switching to a different AI engine that supports a higher token limit, like GPT-4.

From time to time during testing, we encounter extremely long processing. When this happens, we normally see the progress bar for a prolonged period. Then when it finishes loading, the page is blank. This usually happens with slower models like GPT-4, longer articles, or when OpenAI is under strain due to high usage. When this happens, it can cause a timeout (504 Gateway Timeout) error to occur.

Hopefully, this won't happen too often as we've added max_token restrictions back in with v1.2.4, which should limit timeout errors on all models. However, if you do encounter this, we would suggest changing your model to "gpt-3.5-turbo-16k" on the settings page.

An LLM, or Large Language Model, is a sophisticated type of artificial intelligence that excels at understanding and generating human-like text. These models, such as GPT-3, GPT-3.5 Turbo, and GPT-4, are designed to process and manipulate language in a way that makes them incredibly versatile and useful for a wide range of applications. LLMs like GPT-4 are trained on vast amounts of text data, allowing them to comprehend and generate text in a natural and coherent manner. You can think of LLMs as digital wordsmiths capable of crafting content, answering questions, and engaging in conversations with humans. They can be harnessed to automate tasks, streamline content production, and provide valuable insights. Whether you're seeking to create SEO-optimised articles or engaging blog posts, understanding how LLMs like GPT-4 work can help you to generate higher-quality content. For more information, check out our how does ChatGPT work article, which explains all of these concepts in a simple story form.

Intelligence and speed GPT-4 is smarter and can handle longer prompts and conversations without making as many factual errors. On the flip side, GPT-3.5 is faster in generating responses. Model size and parameters GPT-4 has a much larger model size, rumoured to be around 1 trillion parameters, compared to GPT-3's 175 billion parameters. This allows GPT-4 to handle more complex tasks and generate more accurate responses. Training data and contextual understanding GPT-4 has a larger training dataset, which gives it a broader knowledge base and improved contextual understanding. This translates into better accuracy and the ability to handle longer text passages and maintain coherence. Resource requirements and cost The advancements in GPT-4 come at the cost of increased computational power requirements, making it more expensive to run. Tokens and context windows GPT-4 can retain more information from your chats, allowing for improved responses based on your conversation. It can handle around 8K tokens, whereas GPT-3.5-Turbo is limited to 4K tokens. GPT-3.5-Turbo-16K can handle roughly 16K tokens and GPT-4-32K can handle 32K tokens. The latest GPT-4o models that we've added allow for up to 128K tokens.

Tokens are the building blocks that language models like GPT-3 and GPT-4 use to read and process text. A token can be as short as one character or as long as a word. For example, the word "apple" could be a single token. Tokens are not always whole words; they can also be parts of words or even include spaces. As a rule of thumb, one token is roughly equivalent to 4 characters in English text or 0.75 words. For example:

• 100 tokens are about 75 words.
• A sentence with 1-2 lines might take up around 30 tokens.
• A paragraph could be around 100 tokens.
• 1,500 words would be around 2,048 tokens.

• Cost: OpenAI charges based on the number of tokens processed.
• Speed: Generating more tokens requires more time.

It's important to remember that AI Scribe doesn't actually create the content. The plugin defines prompts, rules and guidelines to help instruct the OpenAI model you select. The more refined your prompts and custom instructions, the more likely you will be happy with the content. However, even with the exact same prompts and instructions, you can often get very different results due to the random nature of how LLMs work. If the response doesn't look correct or you are not happy, please follow these suggestions:

• Simply click the "Regenerate" button next to the response or click the "Update settings or prompt" button next to the prompt without making any changes - either option will run the prompt again and the second run may be more to your liking
• Manually update the prompt and then click the "Update settings or prompt" button - sometimes minor changes and even re-ordering the elements of the prompt can make a big difference
• Check and modify the custom instructions prompt on the main settings page

AI Scribe was designed and developed by a UK based digital marketing agency called Opace.

If you don't have an OpenAI account yet, you can sign up for one here: https://beta.openai.com/signup. Log into your OpenAI account. Click your username in the top right hand corner and select “View API keys".Then on the next screen, create or copy an API key.