So let's get to the basics.
Information is saved to storage devices as data. The internet is a network of these devices, allowing one to discover information on another. Data - news articles, technical papers, wikis, directories...
Search engines helped us index the data in these devices, allowing us to find information easily. Google, Bing, DuckDuckGo.
Conceptually, this is all we needed to find or consume information. Then came web and mobile applications. Why? Because it’s not enough for every computer to go to each of these other machines and access raw data. It needs to be converted to a format depending on use cases - banking apps, social networking apps like Facebook or Instagram. Why did we need these apps? Because it’s not easy to store the data and present it to others from our own devices. Each device needed software for each purpose - web servers, databases. Compute was expensive, storage was expensive, and web security was critical. So cloud-based applications made sense. They had the expertise to build apps, and they gave every user a bit of space on their servers. In return, they got paid through ads or subscriptions. This was a fair trade-off for the era.
Until this point, it made perfect sense for why we were storing our data in someone else’s machines (cloud servers).
But now something is fundamentally different. We can all have our own technology expert: AI agents. An AI agent can be our own security expert, our own operating system, our own application. Compute is much cheaper compared to how it used to be, and capacity (storage size, RAM) is much higher on consumer-grade devices. We can save all our life’s data on our own devices (backed up in cloud storage), while businesses like banks and utility providers provide access to our information on their servers to our AI agents.
Shouldn't this be the primary path of evolution for consumer tech, instead of still continuing to rent space from cloud-based AI providers and storing our data in their servers? Dumping our life’s data into their servers isn't a technical necessity - it's a business model. And it's far worse than sharing every bit of our life's events on social media sites. Social media shares our events; cloud AI consumes our identity. It's not just about storage; it's about training. We are feeding our life's data into models that learn from it, turning our personal experiences into their proprietary advantage.
Affordability is the real engineering challenge ahead. Not everyone can afford a $3,000 Mac to run a local model yet. But we need to make progress towards building business model that favours the evolution of consumer ai towards empowering people instead of exploiting their data.
I dream of a world where all my personal data exists only in my own devices (banks and government databases are the exceptions) and my AI agent (it should know whats best for me) decides explicitly who I share it with. This is how it works in our real physical lives.
Tech we use should empower us to be better and more successful in our lives, not just continue to empower billion-dollar companies to become trillion-dollar companies.
Are we building and using AI the right way? I believe it’s not too late to correct the course. but a lot more innovation needs to happen in the open-weight and AI agent arena to bring AI to everyone the right way.