Chat GPT Prompts

Company Letterhead

{{company_name}}

{{company_address}}

Phone: {{phone}}

Email: {{email}}

Website: {{website}}

Introduction

This policy outlines best practices for employees of {{company_name}} when generating prompts for Artificial Intelligence (AI) conversational agents, specifically ChatGPT. Effective prompting ensures the AI provides relevant, accurate, and consistent information, thereby enhancing productivity and maintaining professional standards. Adherence to these guidelines is mandatory for all employees utilising AI tools in their work.

General Principles for Prompt Construction

1. **Clarity and Specificity:** Prompts must be clear, concise, and unambiguous. Avoid vague language. State exactly what information or output is required.

2. **Contextual Information:** Always provide sufficient context. The AI does not have prior knowledge of your project or task unless explicitly stated in the prompt.

3. **Define Persona/Role (Optional but Recommended):** If applicable, instruct the AI to adopt a specific persona (e.g., 'Act as a marketing specialist,' 'You are a customer service representative').

4. **Specify Output Format:** Clearly define the desired output format (e.g., 'List 5 bullet points,' 'Write a paragraph,' 'Provide a table with columns A, B, C').

5. **Constraints and Limitations:** Include any necessary constraints or limitations (e.g., 'Keep response under 200 words,' 'Do not include personal opinions').

6. **Review and Refine:** Always review the AI's output and iterate on your prompts to improve results. Learn from previous interactions.

Prompt Structure Examples

**Basic Prompt:** 'Summarize the Q3 financial report.'

**Improved Prompt (with context and format):** 'You are a financial analyst. Based on the attached Q3 financial report from {{report_date}}, summarise the key highlights, challenges, and opportunities in no more than 250 words. Present the information as three distinct paragraphs.'

**Task-Oriented Prompt:** 'Generate five creative headline options for a new product launch. The product is {{product_name}}, a sustainable {{product_type}}. Focus on benefits related to {{key_benefit_1}} and {{key_benefit_2}}.'

Security and Confidentiality

Employees must **NEVER** input confidential, proprietary, or sensitive company information into ChatGPT or any other public AI tool. This includes, but is not limited to, client data, unreleased product details, financial projections, or personal employee information.

Any information entered into public AI models may be used by the model developer to train their models, potentially exposing company secrets. For sensitive tasks, consult with the IT department for approved internal AI solutions or alternative methods.

Ethical Considerations

1. **Accuracy Verification:** Always verify information generated by AI. Do not present AI-generated content as factual without independent validation.

2. **Bias Awareness:** Be mindful that AI models can perpetuate biases present in their training data. Review outputs for fairness and neutrality.

3. **Plagiarism:** While AI can generate text, it should not be used to plagiarise. Always attribute sources where appropriate and ensure original thought or significant editing is applied to AI-generated drafts.

4. **Appropriate Use:** Use AI tools for their intended purpose – as an assistant for drafting, summarisation, brainstorming, and research. Do not use AI to generate misleading or harmful content.

Training and Support

{{company_name}} will provide training sessions on effective AI prompting techniques and responsible AI usage. Employees are encouraged to attend these sessions and seek guidance from their line managers or the IT department if they have questions or concerns regarding AI tool usage.

Contact Person: {{support_contact_person}}

Email: {{support_email_address}}

Internal Resource Link: {{internal_resource_link}}

Policy Review

This policy will be reviewed annually or as needed to reflect changes in AI technology, company policy, or industry best practices. Employees will be notified of any updates.

Last Revision Date: {{last_revision_date}}

Signature Block

_____________________________

{{authorized_signatory_name}}

{{authorized_signatory_title}}

Date: {{date}}