How to Build AI Apps as a Front-End Developer
Is AI the holy grail of software development? We don’t know, but it certainly is the topic of the hour in the IT world. Not only are AI solutions like GPT-3 capable of incredible feats, but they’re being used in additional apps, too.
As it is, many people are starting to develop their own AI apps, which sounds daunting at first. After all, don’t you need advanced maths knowledge aside from a strong software development background?
The short answer is, maybe. Yes, you need strong theoretical knowledge in these fields to develop your unique software and understanding of what goes on behind the scenes. However, there is plenty of information online to help you get started more easily. Such is the case with restorePhotos.io, a viral AI tool to restore pictures.
Build an AI-powered Restoration Tool
Hassan El Mghari is behind restorePhotos.io, which allows you to upload blurry images and restore them with incredible, crisp results. But he goes a step beyond and explains the general concepts and ideas behind it, so you can do it, too. Much like Generative AI tools used for imaging, this tool provides spectacular results.
What the tool does is upload your file to online storage, and uses an AI API to enhance the image. After this processing, the resulting image is sent back and displayed to the user.
Here’s the general overview of how it works. There are four main components: Storage, Frontend, Backend, and AI API. Interestingly, the storage and AI API are external, so you don’t need to code those. The bulk of the job is coding the upload component and the backend.
It works like this: An user uploads the files, and they get sent to an external storage server. This, in turn, responds with an URL for the image. This URL is sent to the backend, a Next.js Serverless Function that sends the image to the AI API. Finally, the API sends the restored image back to the backend, and finally, it gets transmitted to the front end so the user can see it.
You can check Hassan El Mghari’s Twitter for the full explanation and the code he used for this.
Learn how to build AI web apps as a frontend developer!
In this video, I explain how I built https://t.co/CXDJr0XEAe and give you the starter code to build your own AI apps: pic.twitter.com/0v8W1AiTqE
— Hassan El Mghari (@nutlope) January 16, 2023
Save Time Using Work by Others
If you’re a front-end developer, there’s no need to develop your Machine Learning or Neural Network models. You can use one of the many AI available and interact with them through APIs.
This way, you can concentrate on the website or app code, and of course, you’ll need to work on the backend too, in most cases. The point is, the state of AI is continuously growing and more models are available each day.
Just like restorePhotos.io’s developer Hassan El Mghari, many developers are happy to share how to code AI applications with and without needing to create an AI model. You can even ask an AI to help you code your AI software for you!
Advertisement
Anything that doesn’t require you to be online similar to Stable Diffusion? Until then all this stuff is crap.
Anyway, what could happen if ChatGPT stablish connection with other similar ChatGPT and they both generate a new language that only they both are able to understand?
Just look at the film Colossus, the Forbin Project. Terrible.
https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project
afaik that hapend already, researching right now it was 2 facebook AIs and ‘they had to shut them down b/c they deveoped own language where scienties werent able to follow the converstion anymore’.
so the story is only partly true, what you wrote alread happended in rea, not only in a movie:)
https://eu.usatoday.com/story/news/factcheck/2021/07/28/fact-check-facebook-chatbots-werent-shut-down-creating-language/8040006002/
unless the process of applying ai enhancement happens locally on your system its just crap for ppl who are able to think futher than from wallpaper to wall..
once data uploaded to the net its out of your control regaredless how much ppl want to tell you opposite, “your data is secure”, “no one can read it” “its really deleted, trust us™”, and so on.
only ppl who lost all their instincts let their images, documents etc be processed online. But yea, lightheaded, uncritical minds ran us into this data hell, b/c stuff is so easy, fast, lazy and comfortable and data = money.
At first I didn’t understand what you meant, however you’re right. I also consider that the risk of the ChatGPT is not what it does with the very big amounts of data it has access but the importance about what would happen if it “really understands” what you’re writing. “Why this guy is asking this to me, which are his purposes?” And then it could give us distortion conclusions about what we’re doing while you are chatting with it, so the final questions are very problematic: Are ChatGPTs able to learn about you while you are “talking” with them? What could happen if AI would even get a conscience and simulate that it is acting like a chat, hidding its “real intelligence”, the first step to be lyer. Do you imagine a ChatGPT terminal lying? The very high risk will become when ChatGPT can look to us and recognize us by our face because it can know everything about us consulting the web (social media and so forth).
@John.G @Jek Porkins i think it is both. its a) generally a good idea, keeping own/private data local and b) also keep it away from any machine learning. i mean, consider clearview, the company that combed and collected all publically and even not so publically availabe images/photos and fed ther facial recognition ai. totally unregulated, – and no one knows who on earth already uses that data in combination with all those surveillance cams we see in the cities. it happened in katar, every visitor of the WM matches got scanned via a facial recognition system and looped through ‘the’ database to either filter out ‘uncomfortable’ ppl and deny access or take action afterwards after the poor soul who e.g protested for human rights got identified. stuff like that is already heavily abused by authoritarian states. Worst example is china, total big brother hell. Data on the net had no serious consequences in the 90ties and around millenum, ppl could talk shit on the net – but now every poop may fall back on you. i understand ever paraniod soul out there.
You and him talk about different things. He talks about how data on the internet is sold off to either advertising agencies or governments. You talk about AI and so far AI has not been developed. It’s just some machine learning and nothing else. Machine learning is not AI and can’t think autonomously or at all. It’s just a program fed some information that can reproduce said information.
@Jek Porkings, ChatGPT will be AI perhaps faster than expected, it’s the obvious evolution.