- Promptedia
- Posts
- From Disappointment to Discovery: Prompting AI's New Approach to Naming
From Disappointment to Discovery: Prompting AI's New Approach to Naming
Learn how I transformed AI's arbitrary name scoring into a reliable method by giving ChatGPT a set of measurable characteristics to score, making comparisons more straightforward and easier.
Jim Washok
July 05, 2024
Hey Prompt Masters!
Welcome to part four of this series on engineering AI prompts to brainstorm, evaluate, test, and select a name for a newsletter. As this process is applicable to most any new product or service offering, I want to share with you what my next step was after I felt let down by ChatGPT in the prior step.
If you read part three, you know that AI surprised me with how it arbitrarily scored the four potential names for the newsletter. The resulting ranking differed quite a bit from my name preferences going into the exercise. So, I prompted ChatGPT for a pros and cons analysis to help me select the right name to go with.
That still wasn’t good enough for me. I wanted an apples-to-apples comparison. I needed to take a different approach…
Today’s top topic tags: #productdevelopment #businessstrategy #evaluation #comparison #scoring
Comparing apples to orang… apples
The question I had at this point was how to compare items that don’t have quantitative or other numerical measures that put them on equal footing. Like comparing models of computers or vehicles is easy given that they all have specs with specific values to match. The pros and cons of potential names for an offering don’t have such organic comparables.
Perhaps if I give AI a set of characteristics that pertain to the success of the newsletter and all the names can be assessed against and given a value for a more equitable method of comparison would be possible.
The prompt
Well over a dozen characteristics came to mind that I felt AI could assign metrics for. I asked AI to do just that for each of the four names and put the results in a table that way the values could be summarized, averaged and otherwise ordered and analyzed.
Please create a table with each of the four naming options in a column and each of the following traits as rows and for each name assign a grade for each trait on a scale of 1 to 10 with 10 being the best.
Most of these should be obvious from the row label as to what I'm asking you to rate on. A few I felt you could benefit from explanation that I put in paratheses following the trait name but please don't include the portion in paratheses in the table itself...
Uniqueness (versus the names of competing newsletters)
Ease of spelling
Pronounceability
Memorability
Shareability
Versatility for business purposes (such as future products and services)
Versatility for creative content (such as fun section names & videos)
Professionalism
Creativity
Fun and lightheartedness
Good for AI newsletter
Good for AI digital courses
Good for AI consulting
Good for AI applications
Good for AI company name
Competitiveness (as in ability to attract attention for market share)
Media outlet interest (as in likely to get mentioned / gain coverage)
Coolness Easily marketable / promoted (plenty of creative play options with the name)
Acquistion attractiveness (likelihood to be bought by a media outlet or AI company)
To be sure ChatGPT understood what I meant, what I was looking for or how I was thinking about each one of characteristics to score on, you see that I qualified some of them.
The response
Here's the table with the four naming options and their grades for each trait on a scale of 1 to 10:
Summary
- GNRAIT: Scores well across most traits, particularly in versatility, professionalism, and suitability for various AI-related purposes.
- MAiGNIFY: Highest scores in professionalism and versatility for business purposes, making it a strong contender for a master company name.
- aiquiri: Excels in uniqueness, memorability, and fun but slightly lower in professionalism and suitability for consulting.
- creaided: Has balanced scores but lower in ease of spelling and shareability. However, it scores well for creativity and digital courses.
This table should help you in making an informed decision on the best naming option for your AI business components.
With the scores now determined by AI for each of the nearly twenty measures that I gave AI to evaluate, I could now run a variety of calculations and frequency formulas to see how the four options compare.
Compared to the arbitrary scoring that I had ChatGPT come up with on its own the name aiquiri has fared much better. GNRAIT scored very well but traded places with MAiGNIFY among the top two slots.
Not only did MAiGNIFY beat out GNRAIT this time, this method of prompting and scoring made it possible to do frequency tallies of the individual scores to reveal that MAiGNIFY has more scores of 9 and 10.
What this scoring also showed is aiquiri is not as bad as previously thought. Granted it came in one point lower than GNRAIT yet it landed nearly 40% more scores of 9 and 10 than GNRAIT did. That makes aiquiri just as strong a consideration as the other two top scoring names.
Coming next…
Given that aiquiri scored much better with this method than the prior scoring method and that it was previously my fav for reasons I’ve shared in prior posts, I decided to pursue this option further.
The next step was to engineer prompts for ChatGPT that help me progress on to visualizing what the branding could look like for the newsletter being named aiquiri.
Until next time,
Jim Washok
P.S. Got a cool prompting method you want us to feature? Reply to this email and let me know!
Reply