Self-Sovereign Data and AI/ML: The Dawn of a New Data Economy
With the current boom in artificial intelligence (AI) and large language models, the intersection of self-sovereign identities, data, and AI is becoming increasingly important.
We’re in the midst of an AI revolution, with many experts in the crypto space transitioning into AI gurus. However, today’s discussion isn’t about jumping on the hype train. Instead, we will explore the relationship between self-sovereign identities, data, and AI. This topic is particularly relevant as I’m currently working on a product that combines these elements, so stay tuned for more discussions on machine learning in private data.
New challenges — how to fuel quality data to model
New technologies often bring new challenges. While some fear AI might take over their jobs, I’m confident it will open up many exciting opportunities. One interesting research paper I recently read suggested that even smaller models with fewer skills than large language models, like GPT-4, can yield better results when fed high-quality data. This brings us back to the age-old question: where can we source quality data?
Many startups and companies have the financial resources to train their models, but they need vast data to do so. Where to get this data is a big question. We’ve previously discussed the data economy and how it resembles a Soviet Union-style system. You produce tons of data but don’t benefit from it because your data is locked away on various platforms. This needs to change, and self-sovereign identity (SSI) could be the key to this transformation.
SSI as a key to Data Economy
So, what’s the connection between SSI and AI? We’re already seeing emerging use cases that leverage SSI and zero-knowledge proofs to verify the authenticity of data. With the rise of generative models, distinguishing between generated and authentic content has become challenging.
Authentic sensors — how to proof nongenerative content
We’re now seeing the concept of ‘authenticity sensors’, such as cameras that produce zero-knowledge proofs alongside photos or audio and keyboard input authenticity. These attestations ensure that the information you receive hasn’t been synthesized and was recorded from a natural source.
Another intriguing domain on the horizon is the confirmation that a specific model with certain parameters produces the results you receive. This is a whole new area in the data economy. You can provide your data, trade your data to train models and sign your data as authentic.
ZKML — proof of Model use
However, a big question remains: how can we calculate your contribution to a model’s training? How can you earn income or a percentage of your data or model usage? If your data contributed to the synthetic output of a model and this product was sold, how can you get paid for your contribution?
This is a massive market and a vast business opportunity at the intersection of machine learning and self-sovereign identity. We’re on the cusp of a significant wave in the data economy, raising many questions. For instance, large language models like GPT-4 use books and internet materials without considering rights and privacy.
In conclusion, the age of the new data economy is upon us, and it’s time to start thinking about how we can navigate this landscape responsibly and ethically. The intersection of self-sovereign data and AI/ML is not just a technological revolution; it’s a societal shift that will redefine how we view and handle personal data.
Self-sovereign identity (SSI) is becoming the best friend and prerequisite for the new AI world. We already have a lot of use cases, such as authentic and attested sensors that can prove that audio and photos are organically original and authentic data and data ownership. Zero-knowledge machine learning (zkml) and proof of model use are also emerging as important concepts.
I’m keen to see more and more cases in this space. My favorite one is the data economy — we need protocols for quality data. We need a way to pay back people for data used for models. A big open question that will create new business models is how to measure the contribution of your data to a model and the produced result.
Stay tuned for more discussions on this topic as we explore the future of the data economy and the role of AI and self-sovereign data in shaping this new era.
