Part 1 of How an AI-Enabled Academic Overlay Revives Science
The AI Cat’s Out of the Bag
It’s time to admit despite our misgivings that AI is unstoppable. We must embrace AI and seek to sculpt the adjacent possible we want to live into or risk being overwhelmed by the coming changes. This is a personal cross that we must bear ourselves. For some, the notion of embracing AI may be difficult. But it seems necessary. (More on this in a future post.)
What might embracing AI look like in education? The preponderance of conversations around AI in the educational realm are about students using ChatGPT to write assignments and detecting text written by AI. If we were to embrace AI, we might focus on how AI can do a better job of grading homework—something that sucks the life out of teachers, assessment of where students are on their learning path, and creating personalized learning plans. The conversation could switch to structuring assignments such that the student uses AI to extend the learning opportunity if not prepare them for using AI in their future work. How can the student not only learn, but contribute to building knowledge or fixing things in their community?
I don’t know if it’s true, but I had the thought that one of the primary distinguishers with respect to the winners and losers in this century could be which countries use this AI moment to completely reimagine and restructure their academic, educational, and learning systems for the 21st-century. Education including higher education was broken before AI. Fixing our learning systems now in light of and with the assistance of AI would position young people for the future we are barreling into. The impacts could ripple through society, creating new adjacent possibles that we want to live in.
Embracing AI in Academia
Rather than focus on AI use cases for academia, I am more interested in creating environments (in this case for academia) that integrate AI while prioritizing safety, accountability, and collaboration. In this section, as a submission to the “Reinventing Academic Communication: AI & Academic Writing” dialog, I present the Academic Overweb as a space for the collaboration that leads to writing, a way of ingesting academic content into a connected information ecology, and platform for discussing academic content.
The Academic Overweb is an emergent digital environment for researchers, scholars, and academics operating in the space above the webpage. It includes a meta-community for academics, bots, and an overlay application that is active over every webpage, providing a sovereign layer for academics to connect and collaborate. Currently, participants can access the environment through SDKs loaded on websites and a browser extension. In the future, it will be accessible also through mobile applications and Overweb-enabled browsers. The Academic Overweb incorporates on-chain storage of data and unique decentralized IDs, while enabling people, information, and interactions to have a presence above the webpage.
One of the Academic Overweb’s key benefits is that it creates safe digital space. The environment supports existing credential agencies, enforces one unique decentralized ID per person, and all bots must have an owner, ensuring the integrity and security of academic interactions. On top of this, academics can meet others on the same web page, review their headline, and initiate conversations. This creates the possibility of like-minded researchers meeting while they are thinking about the same thing. AI can help identify people we should meet and potentially augment our conversations. The platform also allows for on-page interactions, meaning academics can leave notes, have conversations, respond to polls, make bridges, label data, and more on relevant web content (e.g., text, a part of an image, a segment of a video) using smart tags. The Academic Overweb uses an attention-triggered design, where web content with smart tags responds to the participant's attention by changing its behavior. For instance, when the participant’s attention approaches content with tags, the content highlights and displays a badge with the number of tags available. The participant's attention is determined by the cursor on laptops and desktops, touch on mobile devices, and line-of-sight in virtual reality. This approach ensures that participants are aware of the metadata relevant to their focus, allowing them to easily access and utilize it.
The Academic Overweb introduces bridges as an essential upgrade to traditional citations (as well as hyperlinks). Bridges are conceptual links between two pieces of online content with a relationship and metadata. Bridges enable deep, precise connections among and between ideas in studies, articles, and the news, not only for citing but also for contradicting and supporting. AI will help identify potential bridges. When a bridge is submitted, the relationship is verified to be correct prior to minting. Bridges self-assemble into the Universal Content Graph (shown as the Collective Learning Map in the above diagram), which provides an online information ecology for academia that connects academic-related information on an idea by idea basis with relationships.
With access to the Universal Content Graph, as one is reviewing a paper, they can see all the related ideas, evidence, and information. Thus, the Universal Content Graph creates a shared contextual view of the relevant information. If something is missing, they can create their own bridge and, once verified, it becomes available for everyone to see. Bridge creators (or owners) earn value based on the value the bridge contributes to the ecosystem and can sell their bridges in the marketplace.
In addition to bridges, the Academic Overweb has a labeling smart tag for attaching metadata labels to pieces of content. This smart tag allows for easier organization and categorization of online content, making it more accessible and useful for researchers and machine learning. AI will help identify data to be labeled. The Academic Overlay also supports developers building their own smart tags and overlay applications that are active over any webpage. Hence, the Academia Overweb can be a sandbox for all academic applications that want to live above the webpage.
Academic writing is often born in collaboration. The Academic Overweb enables communities (e.g., universities, disciplines, projects) to have their own sovereign layer to meet and interact above the webpage. These meta-communities can extend the reach of existing organizations and groups (e.g., the Sociology department at UC Berkeley) to the entire web. The meta-community for the Sociology department at UC Berkeley could in turn nest within the UC Berkeley and Sociology meta-communities, enabling cross-pollination of smart tag content. Meta-communities have their own AI that supports moderation, makes suggestions for people to connect with, and can augment conversations.
With respect to AI Safety, on the Overweb, all algorithms are open while bots and AI content are identified and tied to their owner. Hence, AI content and bots can easily be filtered out. Everyone has their own personal AI for filtering and generation, which stores its data securely on-chain with their private key, and they must grant permission for external use of their data. When hovering above web content, the overlay can display the AI content score for text as well as provenance and chain of custody data embedded in digital assets such as news articles, images, and videos.
The Academic Overweb can not only support deep innovation and collaboration in academia, but also be financially sustainable. The platform generates revenue through partnerships, sponsorships, and by charging fees for premium subscriptions. This model ensures financial sustainability while also ensuring accessibility for all.
Read the second part of the series: Transforming the Research Process