Day 29 of 30 Day AI Challenge – Tapping into the power of Siri AND ChatGPT on your Phone

For those just joining us, every day during this challenge, I’m going to try and do something different or better, using currently available Artificial Intelligence tools.

Today’s challenge is figuring out how to make ChatGPT more accessible when you’re on the move and don’t necessarily have the time to access the web interface. How cool would it be if you could simply ask Siri a question for ChatGPT and let your phone do the rest, promptly delivering a quick and succinct response? You could even do it hands free!

Step 1 – Get the iPhone Shortcuts App

If you know how, you can create some pretty useful routines on your iPhone using the Shortcuts app.

Step 2 – Create a Shortcut to use ChatGPT

There are two ways to go about this step.

The first option – if you’re conversant with a bit of coding you can set up your own shortcut. I’m not a coder but did try and experiment with this route, using ChatGPT to provide the directions and then iteratively troubleshooting.

The second, easier option, is downloading a shortcut created by someone else. There are a few different variations that have been created by other members of the coding community. There are also a couple of great guides by Tom’s Guide and Zdnet available. For my test, I used SiriGPT, which can be downloaded here on an iOS device.

Step 3 – Customize The Shortcut

There’s one necessary step here and a couple of other things you can do to customize how the shortcut works.

Critical – The key thing to understand is that as the shortcut can’t use the web interface of ChatGPT, it needs access to the API. When you create an account with OpenAI, you are also given access to API keys (which you can generate and delete as needed). There is a very minor cost to each query – in my testing this was roughly about 0.5 US cents per query, but pricing can vary depending on which model is in use and your usage behaviour. New users are typically given about $18 in free credit, to be used within a specific time frame (this used to be 3months but could be much shorter now). You can follow the step-by-step visual guide here and copy paste your API key into the shortcut.

Optional – You can do a few of cool things to make the shortcut more accessible and customized to your needs.

• Use Siri to call the shortcut – I tested Siri with the default shortcut name “SiriGPT” but had trouble as Siri kept pulling up web and Wikipedia results. So I renamed the shortcut to just “GPT“. Now, by saying “Hey Siri…GPT” you can run the shortcut hands free.
• Use back tap to call the shortcut – this accessibility tweak is covered on the link provided above. It’s also a handy way to quickly run the Shortcut by either double or triple tapping the back of your phone.
• Customize the voice – you can also select a specific voice used to read out the results that come back from ChatGPT. You’ll see where this can be edited within the shortcut below.

Step 4 – Know when to use Siri, vs when to use ChatGPT

That’s about it. This setup allows you to quickly query ChatGPT when you’re on the run or can’t physically use the phone keyboard.

To make the best use of this though (both in terms of capabilities and any API costs, however low), you should stick with Siri for the things that the iOS assistant typically handles well.

Here’s a handy guide:

What you should be using Siri for:

1. Setting reminders and alarms: Siri integrates well with Apple’s ecosystem, allowing users to easily set reminders and alarms through voice commands.
2. Text messaging: Siri can send, read, and reply to text messages, providing a hands-free communication experience.
3. Navigation: Siri works seamlessly with Apple Maps to provide users with directions and traffic information.
4. Music playback: Siri can search and play music from Apple Music and other supported services using voice commands.
5. Basic web searches: Siri can perform simple web searches to provide users with quick answers to their questions.

Where Siri struggles and ChatGPT can excel:

1. In-depth information: Siri might not be as adept at providing detailed information on complex topics, whereas ChatGPT can generate more comprehensive responses.
2. Contextual understanding: Siri can sometimes misunderstand or lose context in ongoing conversations, while ChatGPT is better at maintaining context and providing relevant answers.
3. Flexibility in phrasing: Siri often requires specific phrasing for commands, whereas ChatGPT can better understand and respond to a wider variety of phrasing and linguistic structures.
4. Open-ended questions: Siri may not be able to provide satisfactory answers to open-ended or opinion-based questions, while ChatGPT can offer more nuanced and detailed responses.
5. Personalized recommendations: ChatGPT can better adapt to user preferences and provide more tailored recommendations, whereas Siri might struggle to offer personalized suggestions beyond its default programming.

Here’s a quick demo:

This should show how both Siri and ChatGPT perform (and vary in their abilities). I asked both to explain the “Fermi Paradox” to me, but assuming I was a 10 year old. As expected, Siri defaulted to a quick web search result, while ChatGPT actually provided something customized to my query.

You won’t hear my voice on the below recording as Siri needs to use the Microphone too, however you can see the results from the first search with Siri, followed by the prompt to ChatGPT and subsequent results.

Till tomorrow…