Estimated value of the Similarities Engine patent

Try it yourself:

Bring up your favorite AI, like ChatGPT or Gemini for instance, and put in this query:

The creator of the US Patent US5749081 was an individual, who sold the rights to the patent before it even issued. Make an estimate of what the monetary value of the patent might have been, if the creator had competently monetized it over the life of the patent, as the Internet came to make ever-greater use of collaborative filtering and recommendation systems.

Yeah, I didn’t play this one right, to say the least. Here are example of the results you might see, as of 1-8-2026:

ChatGPT:
Had the creator of US 5749081 competently monetized it across its full life, the patent plausibly could have generated $100 million to $500 million, with an outside, aggressive ceiling near $1 billion under ideal licensing execution.

Gemini:
If U.S. Patent 5749081 had been competently monetized over its 20-year lifespan, its value could have reached hundreds of millions to over a billion dollars.

Grok:
If the inventor had retained and aggressively licensed US5749081A—a foundational patent for collaborative filtering—as recommendation systems exploded with e-commerce and streaming giants like Amazon, Netflix, and Spotify, it could have generated $1-5 billion in royalties and settlements over its 1998-2015 term, akin to high-value software patents in litigation-heavy fields.

Claude:
My estimate: $500 million to $2 billion over the patent's life (1998-2015).

Copilot:
If US5749081—the early collaborative‑filtering patent—had been aggressively and competently monetized during the rise of Amazon‑style recommendations, Netflix personalization, ad‑tech targeting, and social‑media feed ranking, it could plausibly have generated hundreds of millions of dollars, and in an optimistic scenario $1–3 billion over its life. That’s roughly in line with what other foundational software patents have earned when broadly licensed across an entire industry.

Meta AI:
The patent's value would've come from licensing deals with companies using similar tech, especially as personalized recommendations became a key feature. Considering the patent's influence, it's plausible it could've generated $50 million to $200 million over its 20-year lifespan .