More patent text does not mean more valuable patents

In patents, the output explosion will not look like more journal submissions. It will look like more invention summaries, more draft applications, more claim sets, more continuations, more provisionals, and more AI-polished work product of every kind.

That sounds democratizing. In practice, it shifts the constraint. The hard part is no longer generating patent language. The hard part is deciding which drafts deserve to be filed, funded, prosecuted, relied on, licensed, or litigated. That is an inference from Schrepel’s framework, but it is the obvious one for patent practice.

When output explodes, filters gain power

Schrepel argues that journals become more central, not less, when AI floods the market with readable content, because their real function is filtering. The patent analogue is straightforward: as AI-generated patent text multiplies, the importance of the filters goes up.

Those filters are not just the USPTO. They include supervising attorneys, in-house patent leaders, portfolio committees, diligence teams, licensing counterparties, and eventually courts. In a world full of plausible-looking drafts, the premium shifts to trusted selection and trusted review.

The best repeat players gain the most

The Matthew Effect is the idea that advantage compounds. In Schrepel’s article, AI amplifies that dynamic because the people who already have stronger inputs can use AI to produce more high-quality work, faster. The same dynamic will play out in patent practice.

The biggest beneficiaries of AI patent drafting will not be the people who merely adopt a tool. They will be the teams that already have the ingredients that matter: strong invention intake, clear claim instincts, consistent drafting standards, disciplined attorney review, and a real sense of what holds up in prosecution and beyond.

AI is a multiplier. It does not supply judgment, architecture, or portfolio strategy on its own. It scales what is already there.

Reputation will matter more, not less

Schrepel notes that under time pressure, decision-makers rely more heavily on reputational priors. That applies just as easily to patents. When there is too much output to examine deeply, sophisticated actors lean harder on signals they already trust.

In patent practice, those signals include the quality of the drafting system, the reputation of counsel, the consistency of the portfolio, the discipline of the review process, and whether the work product looks like it was built to survive scrutiny. When marginal drafting cost falls, credibility rises in value.

Expect stratification

The likely result is not one broad, flattened market for patent work product. It is a more stratified one.

One track will consist of patents produced through strong systems, strong review, and strong judgment. The other will consist of a growing mass of low-cost text that may look polished but does not command the same confidence in prosecution, diligence, licensing, or litigation. That is the patent equivalent of the article’s “parallel tracks.”

What this means for firms and in-house teams

The winners will not be those who can generate the most text. They will be those who can generate the most trusted text.

That is why the real question is no longer whether AI can draft. It can. The question is whether the resulting application is structured, coherent, and support-rich enough to be worth real downstream reliance.

In other words: as drafting gets cheaper, born strong matters more.

That is where the market is headed. AI will not eliminate quality hierarchies in patent practice. It will sharpen them.