The Generative AI Layers, Who will win?

This article is a 🌱 Seedling, which means it’s still Work in progress and published in it’s raw form.

More on Seedlings

I have been doing a lot of thinking lately, as the momentum of generative AI continues to accelerate, it becomes increasingly crucial to discern the different strata of players in this transformative arena.

There are essentially three layers of operation to consider:

1. Infrastructure
2. Machine to Machine Middleware
3. Human Computer Interaction Medium

Layer 1: Infrastructure Providers

At the zenith, we find the infrastructure providers such as Amazon AWS, Microsoft Azure, Google Cloud. These companies offer the vital AI infrastructure underpinning cloud applications. In the near future, the democratization of app development may lead us into an era of “Ephemeral Apps”. These transient applications are conceived to serve a singular purpose and then disposed of upon task completion. This shift would inevitably open up a new frontier for the infrastructure providers, essentially making them the gatekeepers of a revolutionary concept, managing the lifecycle of these ephemeral apps.

Moreover, envision a future where Facebook, Twitter, or Snapchat is tailored to your individual preferences, where AI fine-tunes the interface to align with your tastes and needs. The implication of this future trajectory is that traditional SaaS application models might no longer be viable because an AI assistant can now customize solutions to address your specific pain points.

So why is this interesting?
When generative interfaces become a thing, we might move out from traditionally hand-coded HTML, CSS & JS to a set of schematics, like a Declarative Object model that enables interactive interfaces.

Chrome might ship their schematics
Mozilla might ship theirs too

Or an Internet Consortium of some sorts will come up with a set of standards and rubrics for new platforms to use, else we might have have the CMD versus Powershell versus Unix wars all over again, we don’t know, I am not sure.

Less I derail, let me introduce you to layer 2

Layer 2: Machine-to-Machine Middleware

Up next we have the machine-to-machine middleware providers; a.k.a LAYER 2. Current entities in this layer include Hugging Face, Replit, Chroma, Milvus, Pinecone, and Vercel. These organizations effectively provide, simple primitives that developers and app creators can use to build and scale their generative apps. They are an important layer of the stack and in an age of Serverless / On-Demand Computing, it already makes business sense for these provider to offer the primitives and provide ways for app creators / developers to deploy these apps into production environments,

By unlocking access to AI-powered infrastructure, like the marriage of Replit + Google cloud, or Huggingface + AWS — It would be easy for those in Layer 3 to define new models of app design and development that present app developers and creators can easily implement for future generative interfaces.

Layer 3: Human-to-Machine Interfaces

The third layer consists of those who pioneer the human-to-machine interface. Notable contenders here are Apple Vision Pro, Mozilla Firefox, Google Pixels, Google Chrome, and Samsung. These companies represent the mediators, facilitating regular individuals’ interactions with cutting-edge technology.

In an era where generative interfaces may replace traditional HTML & JS with schematics or an Object Model, these companies could have a significant influence on shaping the future of user interfaces. We can already see a precursor to this trend with Apple’s Swift data.

With the advent of these three layers, companies and organizations seeking to bypass the middleware layer would need to get closer to the metal and provide their hardware, a potentially unsustainable proposition as suggested in this article by Moxie Marlinspike. 

Looking to the future, we might witness the emergence of a new age of IoT, fostering a unified hardware interface that all platforms can integrate with. Perhaps we might progress towards an offline > prompt > online world, where our data and interactions reside on a Zero-Proof cryptographic layer accessible via our signatures or private keys.

While infrastructure providers and middleware operators set the stage for the surge in Generative AI, the human-to-machine interface pioneers may be the ones to dictate the rules of the game. They could potentially influence how we interact with this emerging technology, shaping the landscape of user experience, and perhaps even redefining our expectations of personalized digital interaction.