The current development of Artificial Intelligence (AI) and Machine Learning (ML) is fundamentally flawed. Peter Thiel recently described crypto as libertarian and Reid Hoffman describes AI as authoritarian. This speaks to the concentration of power and control that has taken place in AI and ML. Cryptocurrency, as a technology, and Bitcoin in particular, empowers people to own their digital assets, the same way people can own real estate property and intellectual property. The data and information that is needed to fuel and power AI and ML is, however, only owned by a select few firms. They are the only ones who have a meaningful amount of data to truly make an impact and train their algorithms.
The ownership and sourcing of this data is a problem, one that I think crypto can provide a solution for.
Where Does AI Derive Its Source Material?
The first of its two major problems is source material. AI is nothing without the database it draws on. Tencent, Alibaba, Megvii, Sensetime, Facebook, Apple: It doesn’t matter which giant corporation’s tech you’re talking about, they all rely on the input of users’ faces into a massive database in order to function.
This can lead to interesting dichotomies, such as in China. Sensetime and Megvii each provide facial recognition tech to the north and to the south, respectively. Unless people are using both firms’ technology, then each company will only have half of the possible dataset. Due to this limitation, AI becomes not only vertical in its intelligence, but also very regional, reinforcing the very divisions and borders that the internet and networks have been trying to break down.
Who Holds the Power?
Let’s do a thought experiment. Say, one day, Google is able to get the database of every human being and is able to recognize the faces of every nationality and human being in the world, accurately and reliably. Does Google deserve to own that all-seeing AI? What about the people that contributed their faces as data, thus training that algorithm?
This highlights the second major problem with AI — power.
All the power is centralized and controlled by a handful of tech companies. How and why do they have this power? It goes back to the first problem. Without data, and lots of it, your AI is nothing. That means only a few companies that currently exist have true power in AI.
Why should they have complete control and autonomy over users’ data without any input from the public who have contributed so much? Sure, it’s just your face for now — but what about your medical data down the line? It’s a slippery slope to complete control of your life.
In AI, more often than not, and especially in deep learning and machine learning, the technology and algorithms are not the differentiating factors. The differentiating factor is data and how the data set is training the neural nets. There are pockets of data that are highly protected such as our health data. This is the reason why, even today, Google Health has not really progressed, but it could just be a matter of time before the data floodgates are fully open.
Who Should Own Your Data?
In an ideal world, those who have contributed data to train that AI should own a piece of that algorithm and potentially the profits it generates, just as a shareholder who invests their capital can share in its gains. However, unlike purchasing shares in Facebook itself, where, realistically, you have no control or say over how your data is used, having a share in the algorithm itself and, therefore, becoming part of its governance, the user actually has a say in how it’s used.
Owning data and being able to monetize it is not a new idea, but the real urgency is making sure that the crowd owns a piece of the AI algorithms generated by the data so it has input and a level of control. It’s not just about how people can profit from their data, but also how they can be part owners of an AI that redistributes the power structure itself.
It’s about governance but it’s also about the asymmetry of power. It is about taking back control from the large tech firms who, at the moment, run the whole show.
In the future, I see the appearance of vertical data trusts where representatives will be elected to govern the use of different data types for certain uses. I see trusts, the way they are set up today, as the best avenue for these vertical data silos to be governed.
The goal, then, is clear: We need a system that allows us to share our data, profit from it, and crucially, have a say in how it is used. Is there precedent for this? Is there a network where nobody owns the system but anyone can participate in this open, decentralized, borderless, and censorship-resistant network under certain terms and conditions? That precedent does exist, and it has become a part of our daily vernacular: It’s called Bitcoin.
Bitcoin as a Model for AI
Now that AI/data companies have become five of the top-10 companies in the world, those companies will just grow bigger and more powerful. Often behind these successful businesses are AI algorithms trained by data sets. Now that only a few companies own these data sets, it will eventually also be just a few companies that own the most powerful and strategic AIs.
Bitcoin allows for the crowd to co-own and govern a huge digital asset, i.e, bitcoin.
Bitcoin serves as a model for the crowd to share ownership and, importantly, control that ownership. If AI and ML facial recognition algorithms were the same as bitcoin, i.e., public blockchains, that would allow people to be part owners and operators of these powerful and valuable AIs by owning tokens.
How I see this working is quite simple. As a consumer or user, we already “invest” in the AI by contributing our data. Now imagine that data was a bitcoin token, and you can own it the same way as you own a bitcoin token. Then the profits that the AI produces can be paid to the users as tokens. What Bitcoin provides is the crowd ownership of something huge. Think of any AI system or ML system each being a bitcoin that can be borderless, censorship-resistant, neutral and open for anyone that contributes their own private data.
It is not simply about making money or profiteering from your own data; it’s about making sure that data is used responsibly and in line with a majority of the people who contributed.
You would own a piece of the AI software, not the company over it.
This would be a far cry from the lack of accountability in the large public companies who control this data today. Let us not forget that when Facebook was fined for mishandling its users’ data, the company’s share price increased. Accountability among big tech is a pipe dream.
Bitcoin is the first instance where one can own and control a piece of a digital asset through private keys. I’d like to see that happen with data, where people can govern that data and, ideally, get what they deserve for training an AI — a seat at the table and a say in how it’s used. We need to turn these AI and ML algorithms into Bitcoin so that the crowd can own and get what they deserve for contributing — control.
Having A Say in How Data Is Used
This would be an important step for us as a society as we continue to push back against the monopolies and oligarchies that big tech has fostered and allowed to fester over the years. As a society, we need to understand the value of our data, because we can be sure big tech does.
But what about the future? It’s very clear that users will be able to sell their data to companies. However, that's not what’s important and will not solve the fundamental issues. We need to ensure that when contributing data we have a say in how it’s used. We may still disagree with the outcome, but at least we have a voice.
Bitcoin, through the model of true ownership, can bring that: a voice to the individual that’s as loud and clear as the biggest players in the room.
It may seem like a distant dream, but with the power and proliferation of Bitcoin, it’s a lot closer than you might think.
This is an op ed by Phil Chen. Views expressed are entirely his own and do not necessarily reflect those of Bitcoin Magazine or BTC Inc.