Patently Cumbersome

What do NFTs and intellectual property (IP) laws have in common?

I wish I had a good punchline for you, but I’m a data scientist, not a comedian. All I got for you are more buzzwords.

The answer: they are both governance methods for property that is often hard to own, like a brandmark, an image, a sound, a song, or a systems diagram. It got lost in the “right-click-save” discourse of the 2021 boom, but this is exactly why NFTs came onto the scene in 2018, leveraging the power of smart contracts and distributed recording ledger technology to evolve the worlds of IP.

Like most mass movements, as soon as the majority of the population is in on it, the narrative redirects and the revolutionary momentum is lost. (It’s better to find a band before they’re cool than after they’ve sold out.) Hence why, today, many only view NFTs as a get rich quick ponzi scheme. And, for the vast majority of projects in the collective consciousness, it is effectively just that.

After NFTs exploded into the mainstream in February 2021, headlines very quickly faded from artists like Beeple to con artist collectives like the Bored Ape Yacht Club and DeGods, who’ve swindled the masses out of millions while providing little more than a Walmart Fyre Festival — a mini music festival without any of the actual acts, but all of the intoxicants. And while membership passes are a great primitive use case for NFTs, their real power is in transforming data system architectures like the ones that power most of the websites and apps you interact with every day.

That’s where IP rights come in. Many supporters argue that capitalism, innovation, and our most recent innovation cycles — beginning with the industrial revolution — are all fundamentally tied to the idea of ownership rights. Inventors need to be able to own their products, they say, as a way to incentivize innovation and reward those who create new and useful solutions to everyday problems.

I certainly don’t disagree with the premise of that take, nor the clear impact it’s had. My position is simply that, in the age of AI, those laws are archaic, cumbersome, and a bottleneck to the future of problem solving.

The long and short of it is that AI is Pandora’s Box. Sam Altman, in all his eternal wisdom, opened it. There is no deconstructing the replicant LLMs that have been built off of essentially stolen data. But that is also the nature and requirement of the technology to exist — it needs the corpus of human knowledge in its current best state to be its teacher. So now that the LLM irrevocably does exist, what do property rights look like moving forward?

Think of it this way: AI can see all the existing patents, their supporting research, and the spaces in between. That’s how embeddings work. They are mathematical reductions of a concept, and their output is always novel, in the same way that one painting is always the same but each individual person will describe it differently. In other words, AI takes the concept of patent arms race to stalemate levels, all at the expense of the people who would benefit from the solutions.

So while I love the idea of owning an idea, it’s more likely that our future will look like chain-supported provenance, with individuals receiving a bibliographical citation for inspiring the discovery of the creation, rather than full-on ownership of it. As far as paths to monetization go, I personally believe that citation can be sufficient for building a business moat around being the inventor of something novel, while amplifying the other benefit of IP, creating pressure to innovate and improve to stay ahead of worthy competitors.

— El Prof