AI Doesn’t Have to Be Evil (Nontoxic Ways I Actually Use It)
There’s a lot of loud, emotional discourse around AI right now.
Some of it is valid. Some of it is misplaced. A lot of it treats AI like a moral failing instead of what it actually is: a tool layered on top of systems that were already broken.
I don’t think AI is inherently evil. I do think unregulated tech + capitalism + zero safety net is a mess. But that’s not new.
What is new is the opportunity to use AI in ways that are boring, practical, accessibility-driven, and frankly… helpful — especially for small businesses and solo operators who don’t have the budget or time to do everything manually.
Here are two ways I use AI regularly that have nothing to do with replacing creativity, stealing work, or cutting corners.
1. Accessibility checks for websites and logos
Accessibility is one of those things most people want to do right, but the tools available are inconsistent, incomplete, or locked behind paywalls.
There are free accessibility checkers online, but:
They often contradict each other
They focus heavily on code, not design
They struggle with logos, graphics, and visual hierarchy
They miss real-world usability issues
I’ve found that using AI as a second set of eyes is often more reliable.
I’ll input:
A logo
A website screenshot
A color palette
A layout mockup
…and ask very specific questions about:
Color contrast
Readability at small sizes
Icon clarity
Hierarchy and scanning
ADA-related design considerations
This helps ensure design isn’t just pretty, but functional and accessible to more people — including users with low vision, color blindness, or cognitive differences.
AI doesn’t replace human judgment here. It supports it, especially when traditional tools fall short or disagree with each other.
2. Clean, readable transcripts for video content
This one is also about accessibility and it’s a big one.
Accurate transcripts and captions matter for:
Deaf and hard-of-hearing users
Neurodivergent audiences
Non-native speakers
Anyone who prefers reading over listening
And yet… exporting a clean, copy-and-paste transcript from video editors is weirdly hard.
Some tools:
Hide transcript exports behind paywalls
Limit exports
Offer inconsistent formatting
Or just don’t work well at all
The frustrating part?
The computer already knows the words.
So instead of paying extra for something that should be basic infrastructure, I:
Screenshot the transcript view from the editor
Drop it into AI
Ask for a cleaned, readable, well-formatted transcript
In seconds, I get:
Accurate text
Clear paragraph breaks
Corrected punctuation
Something I can actually share or publish
That’s not laziness. That’s removing unnecessary friction and making content more accessible in the process.
This isn’t about replacing people
None of this replaces:
Designers
Writers
Editors
Strategists
What AI actually replaces:
Redundant friction
Paywalls for basic functionality
Inconsistent tools
Time wasted fighting software instead of doing the work
If there’s a more efficient way to do something, especially when it improves accessibility… people are going to use it. History shows us that over and over again.
The real questions aren’t:
“Should individuals stop using AI?”
They’re:
Why isn’t accessibility baked into these tools by default?
Why are basic features locked behind greedy paywalls?
Perhaps most importantly: Why are we blaming users instead of fixing infrastructure? (I am looking at you, legislators.)
AI doesn’t have to be evil. It can be boring. And helpful.
I’m not interested in hype. I’m interested in practical tools that help small businesses do better work without burning out or breaking the bank.
If AI can help make websites more accessible and content more usable without (directly) harming anyone or anything… I’m going to use it. And I think that’s a much more honest conversation than pretending the technology doesn’t exist.
