Why is the output not what I wanted?
updated 7 months ago
In many cases, this is due to improper inputs. Here are some ways you can improve your inputs.
- Be specific. The more specific your input, the better the output.
For example, when describing WordHero, "helps marketers write better and faster" is better than simply "writing tool".
- Include all keywords. Ensure all the important keywords are in your inputs. If you leave out certain important keywords, WordHero may misunderstand what you're trying to achieve - the AI cannot read your mind (yet)!
For example, if describing WordHero, I use "writing tool", WordHero will not know it's AI-powered, so I should use "AI writing tool" instead.
- Spell it out. Unless it's a household name like "Amazon", don't assume WordHero knows a less-known product. Spell it out.
For example, use "WordHero, an AI writing tool" instead of simply "WordHero".
- Avoid terms with multiple of ambiguous meanings. Beware of using such terms. What is obvious to your context may not be obvious to WordHero.
- Avoid niche, industry-specific terms. Beware of using industry-specific terms that have a different meaning in a general context. WordHero may not know what you're referring to.
- Avoid abbreviations. In your inputs, avoid using abbreviations that could have multiple meanings - this confuses WordHero. Only use abbreviations when it's generally well-known and obvious in the context, e.g. "HR", "CRM", "FBI", etc.
- Beware of punctuation, grammar and spelling. Capitalization, grammar and spelling do matter in your inputs, specially if the variants could give rise to multiple meanings. The AI will try to discern whether it's a typo or it's deliberate to give a different meaning.
- Answer the questions as prompted. Read and reread what the inputs require, and look at the example given. Sometimes, users key in something else other than what was asked.
A good rule of thumb is - if you pull a stranger off the sidewalk, tell him your inputs, and he doesn't understand what you're talking about, then the AI probably wouldn't understand as well.
Of course, sometimes there's really nothing wrong with the inputs - the AI is simply not perfect yet.