Content Copyright © 2023 Bloor. All Rights Reserved.
Also posted on: Bloor blogs
Generative AI is “is a form of machine learning that is able to produce text, video, images, and other types of content. ChatGPT, DALL-E, and Bard are examples of generative AI applications that produce text or images based on user-given prompts or dialogue”.
A generative AI (and remember that the actual “intelligence” in AI is very limited) can generate text and high-level computer code is simply a specialised form of text, so it can generate computer code which compiles OK. An obvious question arising from the current generative AI hype, is whether Generative AIs render low code tools obsolete, as the generative AI can just write all your code for you. The short answer is no; for any really significant business application, writing code that compiles isn’t the issue. What matters is to abstract a business outcome opportunity as a coded recipe that compromises total flexibility in the interest of speed and repeatability, and this involves business choices that can only be made by a human. So far, anyway. What the AI can do, is to suggest code, and alternatives, that the human might like to use (and just accepting the AI’s first suggestions is itself a human choice, placing a lot of weight on testing).
The longer answer was explored by Hans de Visser (Chief Product Officer at Mendix), and Amir Piltan (Chief Product Manager, AI, at Mendix) in their webinar entitled “Generative AI: Build Smarter Apps, Build Apps Smarter”. This is the first in a series of webiners, focusing on practical use cases and tools for AI in application development.
To start with, Mendix sees two general uses for AI in the automated business: AI-assisted dev; and AI enhanced business apps. Both are important, and Mendix’ focus on AI is to drive developer productivity – at all stages of the (iterative) dev lifecycle.
If the first generation of low code was heuristic (using prototyping) the second generation uses AI to recommend different choices, the third generation (which we are just entering) uses generative AI to generate apps. The Mendix view of the new lifecycle is bifurcated; you will be able to integrate AI models built in common (non-Mendix) machine learning frameworks into a Mendix app in a low-code way; and you can also develop an AI app using machine learning model training and suitable data (a data scientist effort) and pass it to a Mendix developer for integration with the automated outcome as a whole. Mendix sees the future of business outcome automation as being largely the orchestration of micro apps, each performing a coherent business function.
Mendix definitely thinks that responsible AI matters! Its vision is for secure, compliant and fair use of data and AI and it is currently building AI into its platform.
I don’t think that Generative (or any other kind of) AI will render low code platforms obsolete. But, despite the hype, I am glad that vendors like Mendix are building AI into their products.