ChatGPT: GDPR, data portability, interoperability and the metaverse
One new area of application for sophisticated AI chatbots such as ChatGPT or the You.com search engine, that hasn’t been discussed a lot online, is automated data portability and real-time interoperability and customization, perhaps even a shift from a 2D interface to a 3D « metaverse ».
In my previous job, representing the interests of users in the digital space, I had a vision of a world where users’ data would be fully under their control, hosted on decentralized cloud services, and all of the services they use would simply recombine data of millions of users that have granted access to their data, competing on the basis of their ability to display relevant information with a sleek and nice user interface, rather than relying on the « lock-in » effect by keeping users’ data captive, forcing people to go where everyone else is, which is anti-competitive to say the least.
The emergence of highly sophisticated chatbots such as ChatGPT has tremendous potential to break all of these barriers in multiple ways.
Data portability
The GDPR grants users the right to download their data in a machine readable format, in order to upload it to another online service. But this is rather problematic for a variety of reasons. While your data may be exported in a database language such as JSON or SQL, uploading it elsewhere requires other platforms to be able to “read” the way the data is structured to import it in a correct fashion, ensuring that all of your pictures appear with the correct structure, privacy setting, that all the comments and contents is displayed correctly etc. This is far from being evident, given the highly complex nature of the metadata and how your data is labelled and structured. But using tools such as ChatGPT could automate such a process, serving as an automated bridge between various competing services, translating your data and making it compatible with the import function of another service.
Real-time interoperability
In the future, instead of having to customize and tweak your content depending on the platform or online service you use, the better way of doing things would be to post content directly on your personal decentralized cloud, and use an automated AI to make it available to a series of services, converting your content to make it compatible with the specific format or other limitations of the service. For instance, if you post a lengthy tirade on Facebook, it would convert it into a Twitter thread, or create a summary of your post to fit the character limit. It could manipulate your picture to fit the various format requirements, transform it into an animated video reel, etc.
AI could streamline your content creation and break down “lock-in” effects in a massive way. AI could be used to pull all the contents you have posted in an online store like Amazon, and republish it on another online sales platform at the click of a button. But again, this goes against the GDPR’s data minimization recommendation. Rather than reposting the same contents across multiple platforms, it would be much easier to host all the data directly on your personal decentralized cloud, and let an AI convert it on the fly depending on the requirements of various platforms. These platforms would not host the contents themselves, they would only serve as an interface or a portal to access and view it, streamed directly from your decentralized cloud.
Real-time customization
But the real revolution may come in the disintermediation between online platforms, user generated contents, and users themselves. Since the creation of the Internet, and the exponential increases in data availability online, many solutions have sprung to navigate it effectively. In the very beginning, people relied on giant indexes, which listed websites in various categories. Then came search engines such as Altavista or Metacrawler, which have succumbed to Google. Social media had to tackle the navigation of millions of new user-generated data points, as opposed to the mostly static web 1.0. To that end, sophisticated algorithms were developed to customize a users’ “feed” based on their preference, or rather, based on whatever contents would ensure they would spend the most amount of time on the platform. The next revolution will be for people to use AI directly as a way to not only search, but also access contents on the Internet. Right now, the interface that ChatGPT generates to display contents derived from the data it has analysed is rather dull. It’s a text-based interface with very limited capabilities. The same goes for search engines such as You.com. In the future, I imagine such tools to pull data from all over the Internet, be it from web 1.0 or from your favorite social networks, and generate a personalized interface just for you, displaying the latest tweets, posts, news articles, website updates, videos and other contents. This is really a game changer, since these tools could also be programmed to filter out all kinds of click-bait or sponsored contents. Instead of logging into your Facebook account or your Twitter account, you would rather grant access to these accounts to your favorite AI tool, and it would pull from these platforms contents for you, and rehash it according to your real-time instructions. “Show me inspiring posts” or “show me some updates from friends that I haven’t seen for a while”. Or even “show me something surprising”.
Ultimately, what this means, is the end of platforms as we know them. There will soon be no reason to have a Facebook account in the first place. There are only three conditions which have to be met to completely kill-off current online business models:
1) Ensure that decentralized hosting services are mature enough to handle hosting massive amounts of contents, and also, capable of withstanding the bandwidth necessary to pull/provide access to the data they host;
2) Create a decentralized solution for creating a digital identity and the social links to other digital identities;
3) Develop AIs that are capable of pulling data from decentralized hosting services based on a person’s digital identity, and his social links and ties, and generating an interface for displaying the data on the fly.
With regards to the second point, Vitalik Buterin’s “soul-bound tokens” tick the box of a decentralized digital identity. All that is left, is to create a decentralized service to replace the social links that you may have with other online users. For the moment, services such as Twitter or Facebook not only control your user generated data, but also all the links that tie you to other users, which essentially determines which user-generated data you are “allowed” to see (in the case of private posts) or likely to see (you are more likely to see posts from people you “follow” or subscribe to). These maps and links need to be converted into a decentralized blockchain-based form.
And so now you can start imagining an Internet without formalized “websites”, but rather as contents directly stored on decentralized cloud services, in a machine-readable format, that can be accessed, and recombined by a plethora of various AIs, and recombined for other people to read, mixed with other contents. Of course, some websites and contents would remain, such as websites of official brands and online stores, but ultimately, people wouldn’t really need to visit Amazon. An AI would connect to Amazon and pull from the products the ones that are most relevant to a user, in whatever preferred customized interface.
From 2D to 3D and the metaverse
So far, I have discussed a flat version of the Internet, generated on the fly by AIs. But in the future, AIs could also take the form of a personal assistant, much like the “curator” in the movie “Ready Player One” or the “Librarian” in Neil Stephenson’s novel, “Snow Crash”, which can access the Internet for you and feed you a customized stream of data in a three dimensional form. The flat nature of the Internet creates many hurdles for displaying giant quantities of data, which so far have been solved by infinity scrolling, which has its limits and quickly becomes annoying and redundant. Displaying information and data in three dimensions can make it so much more interactive and especially, enable the display of much more information than on a flat screen. So far, this has mostly taken the form of sci-fi interfaces such as the ones of Tony Stark in the Marvel Universe. But these kinds of interfaces might very well be part of our near future. All it takes is for tools such as ChatGPT to expand into the metaverse, take on the form of virtual humanoid personal assistants, and being capable of generating three dimensional interfaces on the fly, rather than the current textual feedback.
This raises many questions for the future of the Internet, the future of what we thought were Internet “Giants” such as Google or Meta, and also, the kinds of business models emerging from such developments. While OpenAI is busy trying to figure out its business model, many such AI tools are being developed as open source. All of the data that humanity has generated so far… Who does it really belong to? Most of it is a collective and collaborative ownership, much like Wikipedia, which means that ultimately, it belongs to everyone. In the future, we may very well find ourselves in the kind of Internet where people simply cover the bare cost of decentralized computing and hosting, with an occasional charitable donation to pay for maintenance and future developments/improvements of the existing software ecosystem (baring in mind that future AIs will likely be programmed to self-improve, rather than requiring manual updating), much like the Wikipedia foundation model.
How would one advertise in such an ecosystem? How does one “moderate” contents? Perhaps moderation is an obsolete term since no one would really “see” or experience the “raw” Internet (or more specifically, the “raw” World Wide Web), but rather a stitched together version of the Internet based on this or that AI. Perhaps one of the most important issues of the future, will be to ensure that people will freely be able to switch between AIs to access and aggregate content, to avoid one AI to become the sole mediator between users and data. Obviously, people will still have the choice to view the “raw” original content, but as time goes by and these systems improve, only some will choose to view the original content as opposed to getting a precise summary that is specific to what one is looking for.
In any case, it does seem to me that we are at the cusp of a real revolution driven by AI, given the convergence of so many technologies that are coming of age at the same time: decentralized computing, decentralized cloud hosting, blockchain for decentralized identity, social links and payments, AI generated text, image, text-to-speech, video and finally, the hardware enabling a variety of XR (extended reality) experiences.
What a time to be alive!
