While one may think it is entertaining to witness GenAI companies train their AI systems using content from others, these companies may: (1) be intentionally aware, (2) simply don’t care, or (3) even know what training data sets they are using. However, IT MATTERS A LOT!
Let me build a hypothetical scenario: I spent years researching and writing my two books that were published over the period of a decade. Lots of time, money, and sweat went into this. Thousands of other authors have done the same. Now let’s say a startup by the name of GENBOOK gets funded and starts using my book and thousands of other books to train its system to publish new books at the rate of one book a minute. The foundation used for this program was the copyright of books like mine and my publisher’s assets, as GENBOOK has no IP, no assets of its own, and has spent zero dollars on content. It uses perhaps ChatGPT6 or other equivalent products to spew out books in every category. These are not GENBOOK’s property, rather the concepts belong to the millions of assets that GENBOOK accessed illegally to create the only asset its company has, which does not belong to them. And it did so without paying anyone. So, copyright matters; it protects many rights.
Scenario 2: GenAI music startup CHEATERMUSIC is funded. It uses any and all music to train its AI to produce so-called new ‘GenAI’ music. CHEATERMUSIC floods the market with royalty-free and license-free music, built on the hard work of artists and labels, uses ChatGPT by paying a fee, and voila, builds GenAI music via the hard work of others, who are not renumerated. But that is not their IP. It is not their product.
Now imagine you are a tech company. Your IP is manipulated via GenAI companies to do something similar, but not quite the same. With no research or expense, the GenAI company has a competing product. That is why we have fought for patents to protect IP. Now imagine patents disappearing! How much of the technology built by others is improperly used in this way? That is what copyrights do for content, provide protection to those who ‘truly’ built and created things. Patents like copyrights do the same.
Copyright is meant to protect inventors of things: Music, books, films, art, technology, and all other inventions. Today we are seeing a fierce battle between GenAI companies fighting with the might of their large funding rounds, and those trying to protect what is rightfully theirs. An epidemic is brewing and we need to watch it mindfully.
Is it correct to assume a GenAI startup can use other people’s content and assets to create (their products) and generate revenues, while not renumerating those who created it? Do we understand and agree with the far-reaching negative consequences of fighting copyright protection? Check out the latest NY Times battle with OpenAI and many such on going battles. Then ask yourself: Am I ok with this? This is the time for us to have these philosophical and intriguing discussions.