AI and the Status Quo

This is a case for how it will remain the same. AI will certainly change the way we navigate the Internet. But since its consolidation will be a matter of compute power and default consumer touch points, its success will only perpetuate the status quo.

While I’ve often held and shared a degree of skepticism towards emerging tech trends — take web3, virtual reality, or autonomous vehicles, for example — this time, with the raise of AI it’s different.

Unlike with, for example, web3, this time around the big names in tech aren’t just watching from the sidelines, but flooding the space with a boatload of investment and and re-routing their strategic paths to get on the AI train.

They are very interested in seeing this shift happen and we’ve recently seen companies like Microsoft, Google, or Facebook abruptly changing their roadmap to join the AI ride and that’s mainly of two reasons.

First, because the AI domination is a matter of raw compute power and default consumer touch points. Therefore its success will only cement its dominance.

Unlike with web3 or virtual reality, AI’s prominence is aligned with FANGA’s business model because at the end of day, once the technology has been commoditized, it’ll all come down to who can foot the bill for these products — think owners of data centers.

Moreover, consider who already owns the default choices and products that billions of users won’t forsake. Consumer behavior is usually static and gravitates towards familiarity. When somebody is already accustomed to a product — say Google Docs or Microsoft Excel — it is very difficult to make them change and go somewhere else. The consumer benefit has to be insanely better, which it is not often the case. However, you layer a new feature on top of an already known product, give it enough time and value, and eventually users will embrace it.

This presents a challenge for OpenAI, though it’s equally a testament to its accomplishments thus far. A fitting analogy here is Apple Maps, which initially lagged behind Google Maps, but just by the fact of being the default choice on iOS, it so far remains the most used mapping solution for the platform.

Second, because the end game of this AI play looks a lot like a thick foundational infrastructure layer with thin applications built on top.

If this sounds a lot like the Internet as it is today — with thick protocols and big tech riding on top of them — it is because AI won’t create new players like the Internet did, but only perpetuate the status quo. Historically, big tech companies have benefited from Internet’s decentralized, free, and open protocols — such as TCP/IP or SMTP — and created their business models on top. Shaping the Internet as the centralized entity that we all know today.

It then follows that this time around, the ones with the compute power and the data centers won’t only host the party and own the AI models, but also cater to the most common and horizontal use cases. Simultaneously, the entities with niche data will devise vertical applications on thinner layers atop this foundation.

Long story short, I’m not sure if LLMs are part of a technological revolution akin to the Internet. But I wouldn’t underestimate either the power of defaults and the ripple effect of substantial capital infusion in this space. It will undoubtedly shape user behavior and bolster the rule of default products.

First published on May 20, 2023