LEAD's Internal Generative AI Guidelines

2023 has seen a miraculous emergence of the ‘Generative AI Guru’.

Mostly operating through LinkedIn sponsored posts, these gurus fill our feeds with endless long lists of tips and tricks for ChatGPT.

We’ve (unfortunately) read them all and have been on our own journey to get to our own internal LEAD Consulting guidelines for ChatGPT, through a mixture of research, experience and common sense.

We specifically created an internal AI committee, consisting of members of the management team and the technical experts (championed by Joe Whitehead )? – this team now makes the decisions on our AI policies and guidelines, and regularly meet to discuss amendments and improvements. This is a rapidly developing technology, and no one has the answers yet, but that shouldn’t stop us from making a start.

We thought publishing these guidelines, as well as how we’ve got to them, might be useful for other companies too.

?

Step 1: Picking the Right Tool(s)

Generative AI is an umbrella term covering many different types of technology, so we have bucketed the tools into what they produce, i.e., words, pictures, videos. Rather than making an all-encompassing “AI policy”, to begin with we recommend focusing on the most relevant type of Generative AI tool(s) for your business.

How we made our choice:

As a consultancy we create a lot of written content like presentations and reports. We decided to focus on Large Language Models (LLMs) as the first type of AI tool to evaluate.

?

Step 2: Defining Our Why

We’re staunch advocated of not jumping on the Generative AI bandwagon and just use the popular tools because everyone else is using them, so we have been conscious to define the purpose and use cases for how Generative AI can improve our business.

How we made our choice:

After trailing the use of LLMs in some our work we decided that the main benefit would be from the efficiency of articulating our ideas in multiple formats at the same level of quality e.g. whitepapers, presentations, seminars etc.

?

Step 3: Understanding The Risks

Whenever new technology is integrated into our business, we run thorough risk assessments from both a technical and non-technical perspective. We ran an RFP process for Generative AI tools that considered factors from IP risk to team knowledge development risk.

How we made our choice:

The main risk we identified was the handling of sensitive information when interacting with LLMs. We evaluated various LLMs for their data security measures and decided upon ChatGPT as the LLM to adopt.

?

LEAD Guidelines

Our values guide us as we navigate the evolving landscape of AI technologies:

• Excellence: We believe that Generative AI will enhance the quality of work we do
• Transparency: We are making a choice to be open about the usage of ChatGPT as a business
• Collaboration: We are going to help each other in learning how to adopt and integrate Generative AI into our work

?

Tool Guidelines

• Choosing the Right LLM Provider: We have selected ChatGPT for its superior performance and good data controls. However, this choice is subject to periodic review.
• Use of other AI tools: We have not yet reviewed the use of other types of Generative AI. Please reach out to AI committee if you have any thoughts about the use of other Generative AI tools.

?

Data Security Guidelines

• Sensitive Data Protection: Do not enter any information into AI models that could be used to identify a business or individual. Ask AI committee if you are unsure whether your data is sensitive.
• Data Usage: Complete the Opt-out form to prevent your prompts from being used as training data for ChatGPT: User Content Opt Out Request Form.

?

Quality Control Guidelines

• Company Content: Reports, presentations, and blogs must be reviewed by another person before release to ensure they meet LEAD's standards.
• Individual Content: For content such as emails and LinkedIn posts, reviewing before release is a personal choice.
• Content Review Queries: Reach out to AI committee if unsure whether your content needs to be checked.

?

Transparency Guidelines

• Team Discussions: Engage in open internal conversations about AI and its effects on our work.
• Public Updates: We will be keeping our external partners updated with our AI journey through our website blog and LinkedIn page.

?

We encourage everyone to understand and follow these guidelines to ensure the smooth integration of AI into our workflow.