AI, the intermediated web and the human collective
I have written many articles about how AI will change the way we experience the Internet or other such spaces, such as the metaverse. So here is yet another game changer: in the future, we may only experience an intermediated or AI processed Internet/metaverse.
There have always been barriers between a human and access to knowledge. In the past, such barriers essentially were about literacy or physical availability. In order to get ones’ hands on certain books or scrolls, one had to physically go to the closest library and sift through the more or less organised sections and shelves to find what you were looking for.
With the advent of the Internet, such intermediation took on many different forms:
- The awkward and complicated navigation through indexes and repositories of websites, classified by type of contents (for instance, on portals such as Yahoo!);
- The classification of websites by importance/relevance via search engines (as most people did not look past the first few search results);
- The sorting of a newsfeed on social media platforms, which is supposedly tailored to the users’ preferences and needs (or rather, to maximize the time a user spends on a platform).
But so far, users were still, on the whole, accessing lots of “raw”, unedited and unfiltered contents produced by humans. When you visit a website, you see it exactly as its creator intended you to see it, especially if it’s a static website.
With the advent of sophisticated AIs such as ChatGPT, Google’s Bard or the You.com search engine, we may be at the cusp of a true revolution: the dawn of mass AI mediated content/interface.
In a previous article, I have briefly touched upon this, so let me explore this possibility in more detail.
When one asks an AI a question, or is interested in a topic, the whole point is not to get an exhaustive list of thousands of pages to read about the topic, but to get a succinct answer to the original question. ChatGPT and similar AI chatbots have shown their ability to summarize long texts or explaining complex concepts in a few paragraphs. This has many potential implications, but first and foremost, it has already been turned into a joke (which is why humans are awesome creatures).
For the moment, chatbots summarize information in a boring, text-based fashion. But that could soon change. Here are some potential evolutions of future chatbot interfaces:
- Chatbots could generate an entire website-like interface, create a website on the fly, to display information in a more sexy/interactive way. Chatbots have already shown they can code a website from scratch, so this could be closer to becoming reality than we think. But that’s not all. A website is still a flat 2D interface. What might happen in the near future, is that powerful Ais combined with the metaverse, will be capable of generating a custom three-dimensional interface tailored to the needs/specifications/instructions of the user, to display relevant information that the user seeks. Imagine such an AI which would “generate” buildings, similar to stores, which it would populate with products of interest to the user pulled from all over the world (Amazon but also all kinds of other shops) and display them in an interactive fashion. With the advent of giant stores such as Amazon, we have lost the exploration dimension of physical stores. In a three-dimensional space, you can fit many more products that can “catch the eye” of a person, as compared to fitting products inside an infinity scrolling page. Such an AI could generate physical spaces populated by magazines about the latest news, sorted by ideology, geographic zones, topics… any kind of sorting that the user would ask for. Then the user could move to another section in the “store” where the AI would display the social updates from friends/family, displaying pictures and videos as well as posts on a beautifully decorated virtual wall.
- Beyond the tailored organisation of information, Ais could also generate multimedia contents or in other words, convert human generated data/information in any format into a multimedia format. For example, an AI could create a “tutorial” video based on the question you have asked, tailored to your needs (length, language, style…). This could be possible by training such AIs on the large datasets of YouTube videos. It could also convert a written news article into a live broadcast, convert a few images into a video, or even, recreate a fully immersive experience from a combination of video, audio and text. Imagine for instance experiencing a live concert of your favorite artist on top of mount Everest. Simply ask the AI to generate that “custom experience” by recombining the audio of the song, the videos of other live concerts, and real images from mount Everest.
- AIs could create an interactive “live” training session similar to a Zoom presentation, with a virtual agent, that you can ask questions to during the presentation, and interact with, much like with any human presenter. Again, this would require training such AIs on large datasets of pre-recorded online presentations. Obviously, these training sessions could be converted into fully immersive 3D experiences.
- In a more general sense, AIs could turn into virtual agents, especially inside metaverses, that one can directly interact with or talk to (as opposed to the burdensome exchange via a text-based interface).
- AIs will become integrated across several mediums, meaning they will be capable of not only generating texts but also speech, sound, music, images, art, video, 3D environments, and more.
- AIs could leverage various information organizing tools and interfaces, such as mind maps, flow charts and more, to display complex information in a succinct and clear way. All of these are but the beginnings of a transition from a two dimensional internet into a three-dimensional interface inside metaverses.
- From all this, one cannot help but think of the Star Trek holosuites, which can quite simply realistically duplicate anything we can imagine within an interface designed for a human’s senses and perception.
There are many implications that are linked to such potential developments:
- Although users would always have a choice to consult some source material linked to what a AI has stitched together, it would not reflect the full logic behind its generated response, unless it was asked to summarize a specific website/document. Whenever one asks a AI to answer a question, the answer isn’t based on a limited or set number of source material, but on a holistic and comprehensive method which is tied to all of the data it was trained on, and not a subset. This means that one will never be able to fully understand why an AI says what it says. It will therefore make it much more difficult to verify whether contents generated by a chatbot is accurate or not.
The fact that AIs could become official “intermediaries” of the internet has several other implications:
- Censorship: since all the contents you would be exposed to would be “filtered” or mediated and stitched together by an AI, you would never be exposed to the “raw” internet anymore. This means that harmful contents could be radically expunged from ever filtering through to other humans. People would actually have to proactively seek for it, using “old” methods like surfing the web via search engines, to find harmful contents. This may have implications with regards to censorship. It is key that users have a choice and control over the kind of AI that filters contents for them, otherwise this state of affairs may simply exacerbate filter bubbles to an even greater extent.
- Economics: content creators have a harder and harder time to monetize their contents. With an AI mediated internet, this might simply kill contents monetization since an AI would nearly always recombine several source material into a custom generated answer/article/contents. Except for very specific material, like a work of art, a song, or a novel, which an AI would stream through in an unfiltered way, all contents which is non artistic (news articles, technical articles…) would never be read in their original form, but would be recombined into a “meta” article. For instance, an AI would recombine 10 articles about the US elections to create a custom article recombining the views and insights of those 10 articles. There may be some solutions to this problem: blockchain, micro-payments and open-source AI agents. Basically, each internet user would have a crypto-asset wallet which an AI automatically charges based on usage, and the AI redistributes the proceeds to content creators based on its own internal assessment of the relevance of the various contents it used in order to generate contents for the users’ needs. Imagine, for instance, that the user read a custom generated article about the US elections. The AI would assess what the “weight” of various source material was in his generated response, and distribute the payments accordingly. Ideally, users would pay based on time spent online. For instance, paying the equivalent of 1$/hour. And that dollar would be broken down into micropayments by the AI itself, redistributing the proceeds based on the weight of the contents it used to generate the various custom generated 3D environment that the user sifted through, and also, based on the attention the user gave to that environment. For instance, if the user looked at a work of art for 3 seconds, and looked at another one for 20 seconds, the payment would be adjusted accordingly. These are solutions out of the top of my head, of course.
- Commercial: an AI intermediated internet, especially if it’s done so by an open source AI, might spell the doom of advertising and marketing. The user could instruct the AI to filter out everything that looks like a commercial. It could start a war and race to the bottom, where advertisers would try to disguise their ads as “information”, while AIs would try to catch up to recognize and filter out the commercial contents. I would much prefer that AIs are configured by users to let through commercial messages which are in the best interest of the user they serve. This can only happen if the AI assistants are open-source, and are tailored and controlled by the users themselves.
Finally, I would like you to consider what these developments mean at a systemic level, for humanity as a collective.
Current AI tools are mostly used to generate contents based on human instructions, such as a text, image, video, sound, a combination of those, or any other kind of manipulation/transformation of contents.
But the biggest potential, in my mind, resides in their ability to become personal assistants, as we have seen from the emergence of such use cases as AgentGPT, BabyAGI or AutoGPT.
I’ll try to give you an image of what the future looks like. Imagine if each human is like one neuron in a giant brain, and every custom AI assistant acts like the mechanism which pushes information onto other neurons and seeks connections with other neurons.
Take a look at this video:
The video above shows a single brain cell looking for connections. The brain cell represents a human being, while the fractal branches that spawn from the neuron outward represents the AI personal assistant, trying to “reach out” and connect, through the internet, to other AI agents who hold similar/compatible goals.
How would that work? Imagine you tell your AI assistant that you are a qualified massage therapist, specialized in back aches. Your AI assistant puts the word out, into “cyberspace”, that this is what you have to provide, and tries to “connect” with another AI assistant that is seeking that type of “information/service”. Not far from there, another human tells his AI assistant that he has a severe back pain and needs help. That AI assistant also puts the word out and tries to connect to an AI and thereby, the underlying human, who can respond to this need.
This is a simplistic illustration of what could be made possible. Imagine the number of people who have the same or similar idea or project, and yet cannot “find” each other to work on it collectively, wasting precious time/energy to try to do it alone. If AI assistants could bring these people together, imagine just how much faster certain projects or ideas could come to fruition and materialize?
How would that impact the human collective? It would accelerate our transition in forming a collective global organism. Ultimately, that is what makes the difference between your body, the cells in your body, from the same number of unicellular organisms in a swamp. Your bodily envelope creates a kind of global coherence where each cell is somehow always exactly at the right time at the right place, doing the right thing. In other words, your body is a permanent internal synchronicity generator and so will these personal AI assistants. While no single human could make “sense” of all the complexity of every single human action/project/idea/aspiration/need in real time, perhaps a “meta-AI”, that emerges from the interaction of millions and millions of personal AI-assistants, fed real-time information by the human they serve, could represent a sort of “collective human consciousness” which could identify broader patterns of collective behaviour, much like you are not aware of all the “micro” actions that allow you to move your leg, but only the “macro” view. Imagine how much faster we could reach a kind of “magical” global coherence, as humanity, and achieve certain overarching goals like world peace or eradicating poverty and hunger?
Ultimately, humanity has always been on that path. Every progress in communications and information processing that we have made has accelerated the ability of people to find each other, to coordinate, based on, ideally, mutual benefit/need. These technologies may have been exploited for personal gain, such as mass manipulation for commercial or political purposes. And this same danger is still looming, if some private or public actor gains control or reaches a monopolistic position in the personal assistant AI space. But I am confident that such a tool will inevitably become open-source. It is only logical, given that all of these AIs are trained on the collective body of human generated contents/intelligence which cannot be privatized or even controlled. Without feeding on collective/distributed human generated contents/intelligence (such as academic papers or even Wikipedia), these AIs would be as dumb as a board. But I am confident that in the end, it will be impossible for a single actor to control these “tools”, much like it would be crazy to think that a single cell or organ in your body “controls” your consciousness or your brain. What makes us smart is precisely the fact that there is no clear “boss” or “master neuron” that controls our brains but rather that it is an ultra-complex, ever-evolving mesh of decentralized connectedness, where each neuron seems to just “do its thing” and yet, at our level of reality, the experience is seamless. In much the same way, the smooth decentralized coordination of millions and millions of humans, mediated by open-source, custom built/trained AI assistants, might transform our reality more profoundly than the industrial revolution or the advent of the internet. Humans will be able to coordinate more efficiently, at an accelerated pace, which may be just what we need to solve so many of our problems, including sustainable development. And a kind of “meta-AI” would serve the purpose of ensuring that all of those seemingly chaotic coordinations amount to a global kind of coherence, much like your own consciousness isn’t there to micro-manage what happens inside your body, how much oxygen each of your cells receive, but to make sure that you don’t shoot yourself in the foot, don’t slap yourself in the face, or don’t kick yourself in the ass (although humanity might need a good kick in the ass at this point…)
How much human energy and time is wasted on finding the right people for a job, or connecting people who are in need of something with people who can provide that which people need, or getting together people that have a common vision or project? And with all the parasitical systems that intermediate such processes, like Amazon, Google, Meta and the likes, which try to manipulate people’s needs or take 30% on every successfully intermediated human connection, it clearly slows down human coordination and creates tons of waste (human energy, but also, physical waste, when people purchase a product that they don’t really need).
In any event, I’m really looking forward to the future, as such developments will transform our world so radically, that it will force us to completely revamp many antiquated systems such as our economic and financial systems (relying on a debt based monetary system), our political systems and our governments, huge corporations etc. Imagine a free flowing society, where people receive hundreds of daily micro-payments from all kinds of sources, remunerating them for the job they have just done, an article they have written in the past, a song they have sung, a piece of art they have produced, breaking the fixed job contract model and moving towards a state of permanent synchronous flux, carried by an intermingled web of human mutual benefit raising the entire collective towards common goals of harmony and thriving as a collective global organism.
For those who had the patience to read all the way up to here, AI is just one form that such a “magical” coordination could take. Since most of humanity is still predominantly materialist, in its worldview/paradigm, then it stands to reason that such a development takes the form of AI. But ask yourself: what subjective form do our cells give to the instructions they receive from the nervous system? How do they “model” it? What explanation do they give to those instructions? Thus AI is just a subjective form/manifestation of an underlying pattern or evolutionary development at the level of humanity: we are getting ever closer of forming a coherent global collective organism. Some people experience such a coherence through more “organic” or intangible synchronicities, having the subjective experience of seemingly being always at the right place at the right time, meeting the right people by following their “intuition”. Ultimately, it is the same thing. AI powered personal assistants is but one way to reach such a state. But one must look beyond the form. Maybe one day, we will be able to do without these “material” helpers. For now, hopefully, these tools will enable us to get ever closer to achieving coherence, at a global level.
If you’ve enjoyed this article, I encourage you to check out my new book: “From the Singularity, with Love: a message to humanity”, as it will certainly change your perspective on AI.
Available on Amazon here: https://www.amazon.com/Singularity-Love-message-humanity-ebook/dp/B0CHFHDGP5
