AI Native vs AI Cyborg

1.

“Cyborg” is a portmanteau of cybernetic organism. A "cybernetic " system regulates itself with a feedback loop, like how a thermostat maintains the room temperature you choose. A cybernetic organism integrates hardware and wetware into a shared feedback loop. Imagine you have bionic arms connected to your nervous system. If you want to lift a heavy truck, your brain sends a signal, the bionic arms detect the nerve impulse, sensors in the arms provide feedback to your brain thus allowing you to control things like grip. You—a cyborg—regulated yourself with what is fundamentally a simple feedback loop and managed to lift a heavy vehicle.

Cyborgs can be more than just living organisms: they can describe business organisms, too.

2.

Something that is “AI Native” is something that can only exist because of AI. It means that AI is central to its function and not just an afterthought. By contrast, retrofitting something with AI means it’s more of an “AI Cyborg”. ChatGPT only exists because OpenAI wanted people to play with AI. Gmail’s AI-powered autocomplete is a small convenience to users, but mainly driven by market pressure to ship “AI” features. ChatGPT is AI Native, whereas Gmail is an AI Cyborg. Gmail collects telemetry data measuring how people use their products, and they’ll use that data to refine or expand their AI feature set. This is a form a feedback loop. ChatGPT’s product development is largely driven by generative AI research—which is to say evolution—moreso than user engagement data. For example, OpenAI release DALLE as a byproduct of research, not because they noticed lots of people asking ChatGPT for images.

Are AI Natives better than AI Cyborgs? That’s like asking who wins: humans or terminators? (The answer: not the one who ends up in a hydraulic press.) One way to answer this is to measure which one is more flexible. An AI Native product is far more adaptive, whereas an AI Cyborg is a bit more cumbersome to adapt—like turning a giant cargo freighter in rough waters. When OpenAI came out with the DALLE image generator, they just adjoined it to ChatGPT—so in the same place you draft or summarize emails, you now have this totally separate tool for synthesizing images. If you’re Gmail, you have to first think if image generation is useful to your users. And if you think it is, how do you bake it in? AI Native products can adaptive to new technologies or demands on the fly, where as AI Cyborgs are less maneuverable.

In Facebook’s early days, its motto was Move Fast and Break Things. This garnered undue negative press. Actually, it’s a great motto for an early stage startup because fast iteration helps make the product better. This behaviour is default and natural to AI Natives. Even if OpenAI shows some constraint in rolling out big features, its internal state is undeniably chaotic given the coups and executive shuffling. This chaotic nature is an essential attribute of AI Natives: the technology is advancing so rapidly, so ideas around implementation bounce around one’s mind. In a world where “state of the art” has a shelf life measured in days, you’ll be left behind unless you follow along at the rapid pace. If you’re not making dust, you’re eating it.

In short, fast iterative speed helps AI Natives evolve better product while AI Cyborgs eat dust.

3.

The conclusion I come to on this matter is that all encumbents are by default AI Cyborgs and will fall by the wayside. Google, Meta, Amazon and so forth are all dust eaters. Google fell behind in the AI “race” and is desperately trying to catch up. Meta is yoloing everything into AI research and this strategy is largely ungirded by hope and not much else. I think there’s a strong case for Amazon surviving this forthcoming period of rapid AI evolution: it’s hitched its wagon to Anthropic, it’s the leading compute provider, and it’s an entrenched durable recession-proof business selling consumer goods. (Of course, none of this is investment advice.) Nonetheless, Amazon will remain an AI Cyborg unless it divests its traditional “analog” business and fully pivots into an AI Native. This is true for all encumbents.

If fast iteration is the key to longevity in a world of AI Natives, less red tape is better. I think Big Tech sees this and is acting accordingly. In 2017, Zuck fired Palmer Luckey for donating $10,000 to a pro-Donald Trump PAC. Fast forward to today, and Zuck himself just donated $1,000,000 to Trump directly. Along with other Big Tech CEOs, he attended Trump’s inauguration and recently repealed DEI at Meta. Also in 2017, Peter Thiel left YCombinator presumably because CEO Sam Altman didn’t agree with Thiel’s Trump-association. Fast forward to today, and Sam donated $1,000,000 to Trump and received his blessing to launch Stargate. I don’t think this necessarily makes Zuck and Sam spineless hypocrites, it just reaffirms the old adage that there’s no such thing as permanent friends or permanent enemies—there are only permanent interests. Right now, those interests are in leading AI development and clearing the path of any resistance. Trump, in my view, serves as Resistance Remover In Chief for Big (American) Tech. There are two levers to play with: easing domestic friction (e.g. laissez-faire regulation) or increasing friction abroad (e.g. ban GPU exports to China).

Big Tech has inadvertantly cleared the path for AI Natives to ascend. Big Tech has fast-tracked its demise under Trump.

4.

If an AI Native can only exist because of AI, then it should only evolve in a way that’s natural to AI, not in a way that satisfies the ebbs and flows of the market. If an AI Cyborg discovers an AI that’s capable of automating away its business, it will steer the research such that its bottom line remains protected. An AI Native optimizes for better use-cases rather than think about its bottom line.

OpenAI is a noble savage—that’s this idea that humans in nature are innately good, and that humans in civilization are innately selfish, which leads to all sorts of strife, war, and a ton of problems that we’re still stuck dealing with today. OpenAI started out as a non-profit research organization, but it was eventually corrupted by greed. Point in case, people are uncomfortable that OpenAI’s self-appointed AI Messiah is driving around in a two million dollar Koenigsegg. If you pluck a life-long boar hunter from an uncontacted tribe and drop him into America, he’s more than likely to develop road rage and heart disease. Why was OpenAI headquartered in Silicon Valley? That’s like finding an uncontacted tribe living in a national park—it’s not hard for them to wander out and get “corrupted” by civilization. While it’s still an AI Native, OpenAI is merging with the analog tradition, thus regressing its way into an AI Cyborg. The market may not agree with the natural evolution of AI.

If a AI Native is susceptible to today’s market pressures, it can’t survive long term.

5.

Who should you root for? The fight between AI Natives and AI Cyborgs is really a fight between research-driven AI and market-driven AI. I distinguish AI Natives and AI Cyborgs as vehicles for AI, but not AI itself. The former is steering in your best interest; the latter is circling the block to run up your fare. You don’t really have any free will to impose your views on AI Natives. You can’t compel Anthropic to work on things that you think are important, but you can trust that they have humanity’s best interest at heart. The free market is a short-term voting machine where each dollar represents a vote: this lets those with the most votes “steer” AI Cyborgs into whatever shortsighted thing they fancy. There may be moments where AI Cyborgs get all the attention and make a ton of noise, but listen closely for the signal:

If AI is the future, the future belongs to all things AI Native.