TKS Session 5: Ignition Challenge 3
These next few posts are coming a little late to you guys since my website has been down for the past few weeks. But we are back, and you should now see the new blog posts for the sessions that happened previously that I haven’t been able to cover yet.
In this session, it was the final session for the Ignition Challenge with XPRIZE. We focused on how to use AI to create different final deliverables for our challenge project. But, before we get into the tools we used, we started with the MotW, which was done > perfect.
This mindset is one that we come across a lot in TKS, both in the Innovate program and the Activate program. But I understand why it’s such an important mindset. You aren’t able to see results unless you actually finish something and get it out there for the world to see and comment on. The reason it can be hard to adapt to this mindset is that in school, we are told to get it right the first time for assignments or tests. Once you hand something in, that’s final. You can take it back to edit it, and most of the time, you don’t get a second chance after the marking. Lastly, once you hand something in, you probably won’t look at that assignment ever again to try and make it better, especially if it’s something you didn’t care about before.
But if you’re working on articles or projects on topics that you really care about, then it’s more likely that you’re going to publish a first draft, just for the sake of getting it out there. But, as feedback comes in or as you keep learning more, nothing is stopping you from going back and editing to make it better. But if you never put it out there in the beginning or let people see it, you aren’t making progress, and it’s never going to get better.
This mindset aligns a lot with the stuff we do in TKS. In order to make exponential progress in such a short amount of time, you can’t dwell on trying to make every article, video, or project you create as perfect as it can be. If you dwell, you can’t move forward through the focus process or the short timelines for the challenges we do, and then you won’t have anything to show for your time.
It makes sense why we bring up this mindset as we near the end of the XRPIZE challenge. We need to take all the info we collected and learned about for our ideas and turn it into high standards, easy-to-understand deliverables that are comprehensive of everything we learned. That’s why we can use AI to create these deliverables and also keep this mindset in the back of our minds to remind us that this doesn’t need to be perfect, but you need to be able to showcase all the hard work you did and what you learned from research and simulations.
After discussing the mindset, we then went straight into how to use different kinds of AI tools for different kinds of deliberables.
The first thing we looked at was how to use AI tools to create website deliverables. They showed us a couple of new tools beyond the ones we’ve worked with before (like Vercel, Rocket, Replit, etc.). The first one we looked at was Base44, which is very similar to some of the tools I’ve listed, where it’s AI that’s able to create a website for you using prompts. We had done a lot of prompt engineering practice for the past couple of sessions, so it was a good option to work with. The only downside to Base44 is that it’s not as “customizable” in the sense that you can only customize through prompts or maybe small design tools, but you can’t directly edit anything like you would when you’re on a website builder.
The other tool they showed us that supports the option for direct, hands-on website building/customization is Framer. It’s not heavily reliant on AI to create websites, but instead it’s more of a website builder, where it gives you templates/tools to create your own website. So, if you’re more comfortable with website building and using tools such as Figma, then Framer would be a better option.
For our website, we ended up using Vercel to create it.
Next, we looked at how to create prototype deliverables as detailed online replicas and designs. The tool that we looked at is called Meshy, and it uses either word prompts or image references to create 3d models of whatever you’re inputting for it to create. To test it out, I created a couple of basic models of our ClotGuard patch, which is basically similar to the Dexcom diabetes patches that already exist. So, for these models, I had inputted a couple of photos of those Dexcom diabetes patches to see how it could replicate them. Here are some of the models it produced:
The last thing for this session was where our director recorded an AI industry overview video to play for us, showing us how AI could come in handy in the product production/manufacturing phases and what level these AI tools are currently at. We learned a lot about how, although AI is really promising and is able to do some really cool things, there are still some limitations that we may not be able to overcome yet, and there are still things we need to engineer on before we can have it work full-time in areas such as manufacturing, prototyping, design, supply and demand, etc.
And that was it for this session. After this, we would be wrapping up the challenge and moving on with other sessions/project work in preparation for the demo day near the end of this year. This was a really fun challenge to work on and I really can’t wait for more similar to this later in the year!
