How to Talk to AI
- Mike Bogias
- Apr 13, 2023
- 3 min read
Updated: Dec 28, 2024
Recently I discovered this excellent guide on Prompt Engineering for LLMs. Take a look:
What struck me was the parallel between writing prompts for AI tools like ChatGPT and crafting detailed contract clauses to obtain specific deliverables from contractors or consultants.
What's different?
When we write construction contracts, we aim to get the other party to deliver something specific. In our Brave New World, "the other side" is a robot, delivering text, images, presentations, code, and even 3D room designs based on our instructions. While limitations exist on the receiving end, the quality of our communication largely determines the usefulness of the response.
In sports, we learned that a quarterback's job was to "put the ball on the numbers" and lead the receiver. Whether in basketball, hockey, or soccer (football), the same principles apply. The passer bears responsibility for positioning the ball or puck to set up their teammate for success. I see some poetic parallels here with prompts and contracts, though perhaps less dramatic.
AI is likely to produce more consistent results than human contractors or consultants, especially as models become more sophisticated.
In human interactions, text interpretation varies based on education, cultural background, and personal biases—all operating within the realm of honest moral character.
However, we sometimes encounter contractual relationships where language falls prey to "creative interpretation." Some seek advantage through legal systems, testing others' willingness to endure battles over word meanings, endless correspondence, mountains of spreadsheet data, and armies of expensive lawyers narrating their version of the Lord of the Claims saga. In this light, AI may prove more reliable and could even serve as an effective contract arbitrator in the future.
So what's the connection between proper contract preparation and writing effective prompts?
Common tips for contract writing include:
Purpose
What are you trying to accomplish with the clause?
Clear Language
Say what you need. No Jargon.
Specific Language
Leave no room for unwanted interpretations. Even loosely-written P3's, who's flagship feature is "leaving space for innovation" need strategically written details.
Measurable Output
How is the objective met? This is often a pain-point in contract prep involving many stakeholders. Requirements are often written as "for information", which equates to "we didn't price it" on the other end.
Consistency
This applies to both style and structure. Multiple terms with the same meaning add no value. Similarly, mixing specifications based on scope and location creates confusion, leading to duplicates, omissions, and chaos.
Tested Output
Get fresh eyes on your work and verify if your content achieves the desired result. Adjust as needed.
On a related note, last year I obtained a document titled "How to Write an Exceptionally Clear Requirements Document" from QRA Corp, a company developing software solutions for technical writing best practices. You can request the document here.
Now, what about writing a good prompt? Here's what it should include:
Clear Instruction
Define a specific task or instruction you want the model to perform.
Context
This can involve external information or additional context that can steer the model to better responses.
Input Data
Is this an input or question that we want to find a response for?
Output Indicator
Indicates the type or format of the output.
What do they have in common?
Everything.
Purpose, specifics, context, clarity, consistency, measured output, and testing all enhance the likelihood of creating prompts that generate useful outputs. Contract writing skills transfer naturally to the world of prompting. Success requires only a curious mind, strong language skills, and the persistent inquisitiveness of a five-year-old.
The title "Prompt Engineer" has recently emerged. While it may not demand the mathematical rigor we typically associate with engineering, the field is flourishing and can command salaries up to $350K (USD).
According to Delcore, prompt engineering is an artistic way of communicating with AI algorithms by using a mix of technical skills. However, no coding language is required, as prompt engineers simply need a decent level of language and grammar skills, data analysis, and critical thinking.
Could you be a Prompt Engineer? Based on these requirements, many could qualify. Looking deeper, we might find that those with philosophical backgrounds and Socratic reasoning skills are exactly what these positions demand.
I think, therefore Ai?
Some AI Tools to Check Out:
https://www.decktopus.com/ Text-to-Presentation Generator
https://docu-talk.com/ Ai Document Queries (Includes Source Info)
https://www.spellbook.legal/chat-gpt-for-law Ai Contracts-Builder
https://videohighlight.com/ Summarize Youtube Videos with URL link only
https://gamma.app/ Text-to-Presentation Generator
Prompting cheatsheet
If you found this article helpful, please like and share. Also, follow us on Twitter @BluelineProAI.
Disclosures
- Article written by Mike Bogias - Article editing and image generation by AI Software - Image prompts and GIF recordings on Twitter by Blueline Consulting - Published by Blueline Consulting
