Some Practical AI Tools You Can Use Today

To say that tools based on AI text and image generation are having a moment would be a massive understatement. Every app maker out there is seeing how they can add OpenAI’s GTP model (which powers ChatGPT) to their software, and every tech pundit is drifting towards the extreme ends of the spectrum on how impactful this technology is (it’s either the end of the world or it’s glorified autocomplete that lies to you).

While those bigger conversations should be had, today we’re going to look at some more down-to-earth ways that text and image generators can be used today to accomplish useful tasks.

Fake Data

Tools used: Raycast AI, ChatGPT, or any app with OpenAI text generation.

At work, I often find myself creating demos of products, and I need to use fake data on a form to show how something works. Random name and address generators exist, but I often find that these are too random and don’t feel as real as I’d like. Text generators are kind of incredible here, though.

I used Raycast AI (currently in beta) to ask for “5 fake addresses for people in the suburbs of Chicago” and this is what it gave me:

This is vaguely where I live as well, and I can confirm that these addresses sure look a lot more realistic than the completely random things you would likely get for your average random data tool.

Now, there are some limits based on the security and privacy features built into OpenAI’s model. For example, I asked it to give me fake phone numbers for these fake Chicago addresses, and it refused. I could ask it for fake emails, though.

And this goes well beyond just names and addresses — you can also brainstorm company names, usernames, or basically any other sort of fake data you might want.

Clip Art

Tools used: Midjourney

Along the same lines, I like to use Midjourney to create custom “clip art” for presentations that I give at work. Previously, I would do a Google image search for what I wanted and then copy/paste something close into my presentation. Midjourney (or DiffusionBee if you want to render these locally) has been a game changer for getting these little images that help spruce up presentations.

One example use case is generating a fake company logo. I prompted Midjourney with “neon blue and purple logo for an ecommerce company white background,” and it gave me a few decent options.

Obviously, none of these are good enough for an actual company, but they’re decent enough for demo purposes.

Alternatively, maybe you want a rocket showing your company’s potential growth, so you ask for a cartoon version of a rocket in green to match your company’s brand.

Again, not perfect, but totally usable for an internal presentation where perfection is not required.

Wallpapers

Tools used: Midjourney and Pixelmator Pro.

One more thing on the image generation side is the ability to create your own custom wallpapers. This isn’t as easy as “enter prompt, get wallpaper,” but it’s not too far off.

In this case, let’s make a soothing wallpaper with cartoon clouds on a blue sky. The prompt I gave Midjourney was, “phone wallpaper fluffy clouds in blue sky cartoon style –v 5 –ar 9:17” which gave me a few options, and I chose to save the best one.

Quick side note, Midjourney has a lot of parameters you can give it to customize your output, and the two parameters I used above make it use its latest v5 model, as well as making the aspect ratio 9:16, which is generally good for smartphone wallpapers.

The resolution of these images is a bit lower than you would like for a modern smartphone, so you can take the image you downloaded from Midjourney and use an app like Pixelmator Pro or Photomator to increase the resolution by 3x.

At that point, you can save the image to your phone and have something that looks like this:

Of course, there are many artists out there who are creating beautiful wallpapers as well. These image generation tools should ideally be considered additive to the variety you can get today, and not a replacement. If you’d like to check out some cool wallpapers made by people, then BasicAppleGuy, Oliur, Leo Keller, and Shevon Salmon are great places to look.

Explaining Code

Tools used: ChatGPT or Raycast AI

If you are like me and are nerdy enough to need to do some light coding from time to time, but you’re not a pro developer and find yourself Googling how to do things all the time, then these AI tools can come in pretty darn handy in terms of explaining what specific code is doing in an understandable way.

For example, I want to validate an email address is formatted right on my page, but I don’t know how to do that. The top Stack Overflow response seems helpful, but the answer seems to assume more knowledge of Javascript and regular expressions than I personally have. No serious problem anymore, as ChatGPT can explain this in a way that improves my understanding of what code I’m copy/pasting into my project.

There is quite a bit happening in this function, so it’s not easy to share the full response, but here’s the bit that decodes the regular expression into a format that I can understand more clearly.

Some people are worried about the dangers of asking apps like ChatGPT to write code for them and use that code in their software without knowing how it works, but the other side of this is that ChatGPT can be used as a tool to better understand the code that people are already copy/pasting from other humans.

Writing AppleScripts

Tools used: ChatGPT or Raycast AI and Script Editor (built into macOS).

On the other end of the spectrum, we can have ChatGPT (again, or any app with an OpenAI integration) write us a basic script to run on our computer. You can do basically any language you want, but let’s try something that’s useful on the Mac, but most people have never learned how to use: AppleScript.

When I end work, I often want to close my browser, but save a list of the pages I had open to make a task for myself to check the next day. I don’t know how to write this as a script, but I know it should be possible, so I just asked ChatGPT to do it for me. my prompt was, “write an applescript that collects the title and url for all open tabs in Safari and saves them as a Markdown list to my clipboard,” which explains what I want in plain English, and this is what it gave me:

When I ran this script with four tabs open in Safari and pasted the result into Todoist, which has nice Markdown support, this is what it looked like:

And you can iterate on the script if you want! Maybe I use Things, which supports limited Markdown, but links look ugly, so simplifying the output could make it better. This is easy to do, and I gave ChatGPT a follow-up request, “return the urls in the format “{{title}} – {{url}}”.” It happily changed the script to reformat the output to my new requirements.

Perfect! Now I can add this AppleScript to something like Shortcuts or Keyboard Maestro to run it whenever I want.

Editing My Writing

While I have not used any of these tools to tweak this article, I have used them to help me get through some other things I’ve written. The current technology available doesn’t do a great job of being specific or sounding all that engaged, so while you can ask these bots to write whole things for you, I don’t think that’s a useful feature unless you are a content farm that just needs stuff to post.

Heck, I asked Raycast AI (which uses OpenAI) to write this post!

Again, what it gave me was lifeless and non-specific, and you wouldn’t have gotten much from it if that’s what we published instead. So no, this shouldn’t be used to write for you, but it can help you improve your writing in some cases.

As an example, I was writing a blog post and I wasn’t sure what to say in the conclusion. Everything I was writing didn’t feel quite right and I felt stuck. So I just selected the paragraph that wasn’t working, opened Raycast AI, and asked it to “improve writing.” In a couple seconds I had my message reworded a bit, and it was more clearly expressing what I wanted to say. Once again, it was a bit flat and not in my voice, but I was able to make some more edits that got me to a place I was happy to share with the world.

This was a great example of how these tools can be used to augment our work, and not replace us entirely.

Takeaway

The big question on everyone’s mind these days is how impactful these AI tools are going to be in the future. Will we have six months of fun with them and then realize they’re not actually that useful? Will they continue to advance at incredible rates and change society in a substantial way? No one can answer those questions with absolute confidence.

What I find useful when there is a new technology getting lots of hype is to use the tech and see what it’s actually good for today. In my opinion, the fact that there are some real use cases that I’m actually using on a daily basis is really encouraging that there is something of value here.