AI Everywhere : how many geniuses can there really be ?
2024-02-28
“You go online and there you see every add-on, app, website is now offering you to try their new AI solution, well AI is the new thing but they can’t all be machine learning experts overnight can they ? like how come there are so many, is there some quick “framework” they must be using ?”
A dear friend of mine asked me all confused with too many AI apps showing up in his screen everyday and in a logical sense his conclusion wasn’t entirely wrong, there is obviously some magical ingredients that is allowing the marketplace to be filled at this super fast rate and a framework is indeed one piece of the puzzle.
Enter the big three : Framework | API | SDK
Picture this, a developer has a project, where there will be buttons that change color when the cursor hovers over them, since they all act the same you can easily imagine that coding more than one button from scratch can become repetitive and boring very fast, so to solve this he codes one button <a component> and now every-time he needs that button somewhere he just references the component and it becomes like building with Lego blocks instead of a boring repetitive task, now add more Legos and you end up creating a <framework>.
Framework : “a structure that you can build software on. It serves as a foundation, so you’re not starting entirely from scratch. Frameworks are typically associated with a specific programming language and are suited to different types of tasks” —— Codecademy Team
Now this developer created a very smart Lego block that can tell the weather in text form, on a new different project he wants to have the same function but he can’t really reference that Lego block since it’s in an entirely different project, to solve this he edits the weather block so now it can send that weather text outside of it’s home project through a line of connection like a channel signal that the developer can now intercept and manipulate in the new project, so it becomes like a wireless Lego block, and with that he has created and used an <API : Application Programming Interface>.
API : “is a set of rules or protocols that let software applications communicate with each other to exchange data, features and functionality”— — IBM
After a lot of work on the project it just became so full of amazing complex and smart Legos and it had the potential to maybe be repurposed for other new projects, and at a certain level it just became too many APIs to keep up with and write for each block, so the developer decided to package the entire project into one big Lego piece that he can then place and use in other projects and interact with directly to build new and different projects based on that original project, and with that he has created an <SDK : Software Development Kit >.
SDK : “is a set of software tools and programs provided by hardware and software vendors that developers can use to build applications for specific platforms. SDKs help developers easily integrate their apps with a vendor’s services” — TechTarget
AI is the smartest Lego block
So now following the Lego analogy, it should be easier to answer the question about AI being everywhere, well we can safely assume that most of the new AI apps that keep on showing up are a 100% utilizing famous pre-built AI software solutions like GPT-4 from Open-AI or Google AI.
As an example a new AI competitor will most likely be using one of the famous deep learning frameworks like TensorFlow or PyTorch in Python since they are specifically built to help developers train AI models on large data and better utilize hardware resources, now after producing a great solution they will provide an API and/or an SDK to other developers to be able to utilize their product in new ways and build upon it.
Now let's imagine a company that never worked on AI before or has no machine learning background what so ever decides to provide a new AI story writing app to join the AI market, the logical approach for them is to use a third party API , the programming logic would be something like this :
1- The app takes your input and sends it through an API to the third party software
2- The third party does all the heavy lifting generating and sending back an answer
3- The app prints out the answer after adding up some logic of it’s own to make it unique
So it is technically a new AI app, but in reality the number of actual AI solutions is still the same. Of course this is not to say that this company didn’t provide anything different or new, using an API still requires developing innovative ways to best use the provided data and technology, but this approach would explain the rapid rate in which these “AI apps” are appearing at.
At the end of the day we can describe software as a stack of solutions built upon one another and that of course is what allows for new technologies and innovation to emerge, and if it were for people to start from scratch every time we would never have seen this crazy rapid development that we are seeing today.
Hopefully you found this article helpful, thank you for reading.