Ask your task better: writing prompts for conversational Generative AI

The digital realm is a whirlwind of innovations, and riding high among them is conversational Generative AI. As with any technology, the tool is only as good as its user. In the case of Generative AI, this means the importance of posing the right questions or, to be precise, writing effective prompts.

1. Understand Your AI’s Capabilities

Navigating the intricate landscape of Generative AI reveals its paradoxical nature. Its vast knowledge base is commendable, giving us insights at our fingertips. However, it's essential to recognize its confines. AI operates on its training, deriving answers from a set pool of data.

Example: If you prompt the AI about a recent event post its last update, it might not provide current specifics but will offer general context based on its last training.

Dos:

• Familiarize yourself with the AI's training cut-off.
• Ask questions within the AI's known timeline.

Don'ts:

• Expect real-time updates.

• Assume the AI is omnipotent.

• Be Specific and Direct

Ambiguity might lend charm to poetry, but with AI, clarity is the key. The distinction between a vague and direct question can often dictate the response's quality.

Example: "Tell me about climate change?" might offer a broad overview. In contrast, "What are the causes and effects of climate change in the 21st century?" will guide the AI to provide a more tailored response.

Dos:

• Use specific keywords.
• Provide clear context when necessary.

Don'ts:

• Use open-ended statements.

• Expect the AI to always discern the crux of vague questions.

• Context Matters

Think of context as a compass for AI. Without it, the AI might go off tangent, producing results that might not align with your expectations.

Example: If you start by discussing Shakespeare and suddenly ask, "What's his most famous work?", the AI will know you're referring to Shakespeare. However, without prior context, a prompt like "What's his most famous work?" could refer to any male figure in history.

Dos:

• Maintain a conversational flow.
• Reference back to prior subjects when necessary.

Don'ts:

• Jump between topics without clear transitions.

• Assume the AI always remembers long past prompts.

• Experiment and Refine

Perfection with AI prompts is an iterative process. The nuances of language and specifics of a question can make a vast difference in the response.

Example: If "Describe solar energy?" fetches a too-basic answer, refining it to "What are the technical advancements in solar energy in the past decade?" can narrow it down.

Dos:

• Test different phrasing structures.
• Seek feedback from peers on your prompts.

Don'ts:

• Settle on the first response.

• Get discouraged by non-ideal answers.

• Safety and Ethical Considerations:

The vastness of digital information brings a mix of facts and biases. Generative AI, despite its brilliance, isn't foolproof. It's crucial to handle its outputs with a discerning mind.

Example: If you ask, "What are stereotypes about a certain culture?", the AI might provide them. But always contextualize that these are stereotypes and not universally accepted truths.

Dos:

• Verify critical data from multiple sources.
• Frame questions to avoid promoting harmful stereotypes.

Don'ts:

• Use AI-generated data as the sole basis for sensitive decisions.
• Prompt the AI for potentially harmful or malicious content.

Some examples of prompts written better

Below is a Markdown table containing 20 entries showing the expected output, followed by a poorly typed prompt and the corrected prompt:

This table shows how to refine their prompts for more accurate results when using conversational AI.

Conclusion:
In the expansive realm of digital interactions, AI emerges as a game-changer. Its potential, though, is harnessed not just by interacting, but by refining our art of interaction. Crafting better prompts is not just a skill; it's the key to unlocking AI's treasure trove of insights.