AI Policy

Naturally Social’s use of AI in work output has not been introduced to reduce staffing numbers or replace human staff with artificial ones. Any tools being used have been explored, vetted, and approved by our management team and are there to complement the work that we’re creating. While these systems are improving our productivity overall, without having humans with the knowledge and experience about what makes good content to review it, they wouldn’t benefit the team at Naturally Social in any way.

Definition of AI and AI-Generated Content

Artificial intelligence (AI), the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.”
(Britannica Definition).

AI-generated content is any content you can produce using AI. It’s usually writing or imagery. We’ll focus on written material in this blog, although some of our points apply to both content types. Written content can be blogs, articles, social media, web copy, etc.
(Proofed.com).

The AI systems available include (but are not limited to) systems to help with visual content by creating full images and/or videos or by being included within tools like Canva. ChatGPT can be used to create written content, for example, blogs, social media content, essays, etc.

Like with most organisations, Naturally Social has been using AI in the form of Grammarly for a number of years.

Data Input:

Naturally Social recognises and understands that all algorithms are taught, and so unethical bias and prejudices may exist within any training data used.

Information that could allow AI tools to become biased will not be used within any prompt inputted by Naturally Social.

To mitigate any possibility of bias, all content is to be reviewed thoroughly before it is used to contribute to work.

The following information is never inputted into AI systems:

• Personal information (following GDPR processes). This includes information relating to a client/client’s customers as well as Naturally Social team members and private and confidential information.
• Protected information. This is information that is covered in policies relating to discrimination. Naturally Social users will not input information that relates to the ethnicity of the audience they are targeting/creating content for. Other protected categories can be found within our equal opportunities and diversity policy.

Output Review:

When AI has produced output, our team will review it before publishing.

We will be looking for signs of the following:

• Use of correct language (British English, unless otherwise required).
• Use of correct tone.
• Does the content include any derogatory words, phrases, or insinuations?
• Does the content follow best practice and strategic goals?
• Does the content reference any specific protected information?
• Plagiarism.

In the event that any output contains harmful information, it will be raised to the senior leadership team immediately, and appropriate action will be agreed upon and taken.

Our Intended Use of AI:

Naturally Social plans to use AI to contribute to the work we produce for both us and our clients.

In all instances, real photos are our go-to source of visuals. If clients are lacking new imagery, we will, in the first instance, seek to book time to collect updated visuals and video content.

Social media platforms, starting with TikTok, are beginning to put in requirements that content created by AI (referring to video and imagery) be tagged accordingly. Naturally Social is committed to providing full transparency on our use of AI, and a label/badge will be added to the image/video as well as using any provided tags when posting and hashtags if required.