🌻 Edition 28: Monetising The Value in Your Data 💰

Data (quality and quantity) is the ammunition powering the arms race in tech. Stretching from the big data era to LLMs – data is just as important as the models and GPUs in powering the AI revolution.

The nature of IP in the era of the foundational model is of profound significance to both traditional media companies (publishers, brand owners, content providers) and to technology companies.

To what extent do generative models create new outputs, and to what extent do they merely copy? If the generated content is an exact match or near exact match, the argument for copyright infringement becomes difficult to ignore.

Legally, the issue of transformative use vis-a-vis transformer architecture will be resolved through the courts and via the statute book.

Technically, a lot of the copyright issues with LLMs can be avoided through "temperature" - or randomness settings - which control the variance of output.

The issues faced here are not new. Law and Technology have always had a difficult relationship - technology disrupts, and law protects rights - but the relationship is usually resolved for mutual benefit, albeit after much soul-searching and hand-holding.

Over the last decade, we have seen commercial models bridge the gap between technology innovation and legal copyright in the media arena - with YouTube now becoming a powerhouse in the ecosystem, creating a new platform for the global promotion of high-quality brand IP from established players,  while also creating new tools to detect, protect and monetize IP infringement, and critically, skilfully enabling a whole new creator class at the grassroots level, producing previously unimaginable amounts of semi-pro content. So, as it was with content, it will now be with data.

New marketplaces and models will emerge – full disclosure, the author as an entrepreneur is building such a data marketplace – that will enable data scientists, AI researchers, and commercial operators to license data from a variety of qualified, quality sources providing the scale and specific niches they need, while also enabling data providers to monetize their value in ways which respect their value add.

There are legitimate concerns as to the future – “information wants to be free” is a rallying cry for the libertarian audience, while “respect our rights” seems to be the theme driving moral and commercial outrage from the IP holders whose data has been used - without their consent and for no commercial reward. A compromise needs to be found; one that enables the AI evolution to proceed at pace and scale, empowering both smaller research labs and large global players to build out next-generation tools without prohibitive financial barriers, but one that also provides meaningful commercial upside to the data sources powering these new tools, often with claimed dilutive harm to created IP.

The worst outcome would be data balkanization, commercial entities exclusively licensing data by LLM with some kind of content-exclusivity-by-VOD-service platform model that meets the Mexican stand-off patent portfolio approach. Such dynamics would slow down innovation, providing an opportunity for out-of-jurisdiction or unethical actors to simply ignore licensing and train AI agents with greater potency.

The best outcome would be an accelerative, augmented model where licensing deals enable AI researchers to acquire quality data, at scale, from diverse sources to power a multiplicity of sources, in ways that lead to a flourishing of ideas, models, and services bringing about an abundance in human society – a new renaissance era. Data holders would share with a range of commercial models, upfront license fees, blanket deals, and revenue share and could see data licensing create new internal business units to actively create and curate their data for 3rd party monetization. Creative strategies could see smart IP owners get ahead of the game and introduce fan-art packages on a paid basis, selling the outputs as NFTs.

The challenge is set. Do we end up with degraded user experiences - think GDPR on the web or fragmented experiences aka content exclusivity on VOD services – end up in decades-long legal quagmires and a civilizational handover of technology leadership to non-Western, non-IP respecting entities, or do we do as we have always done - adapt, innovate, overcome - to thrive in an era of new possibility?

**

I will publish the next Edition on Sunday.

This is the 28th Edition, If you have any feedback please don’t hesitate to share it with me, And if you love my work, do share it with your colleagues.

It takes time to research and document it - Please be a paid subscriber and support my work.

Cheers!!

Raahul

**