Introduction
You probably heard that now you can call ChatGPT over the phone. This seems unreal. It is cool but do you have an idea whom you’re gonna be talking to? The best you can do is to rely on service provider reputation – OpenAI OpCo, LLC.
Issues with Current AI
- Centralized Models: Most widely-used models like GPT-4 or Claude are closed-source and centralized.
- Dependence on Centralized Infrastructure: Even “open-source” models often run on centralized platforms like Groq or Together AI.
- Opaque Operations: It is impossible to verify how agents operate—including prompts, limitations, and stop lists.
- Data Privacy Concerns: Users have little visibility into how their data is handled or stored.
These challenges emphasize the need for responsible AI use and caution when relying on AI agents for critical decisions.
AI and Learning
Lucky you if you graduated from school before getting spoiled by AI.
But what if you are not able to verify stuff just because you have not developed learning skills, critical thinking, reasoning skills or just because you are too young or you got addicted to “helpful and efficient” AI agents? This could be a serious issue. Calculators began as costly, limited tools, giving schools years to adapt. Now, AI is transforming education instantly, impacting every subject. Students will cheat with AI but also integrate it into their work, challenging educators and questioning the relevance of traditional assignments.
Common understanding is that unless assignments are completed in class, it’s impossible to reliably determine if the work is human-made.
Verifiable AI can change this.
NEAR AI Hub
NEAR.AI is the answer.
While in Alpha, some exciting features are already available to try. Nothing compares to learning by trying so let’s do it.
Go to the NEAR AI Research Hub.
You may sign in with your NEAR account to interact with the platform. At the Agents tab you find all the agents added to the Registry – Decentralized storage with support of private and encrypted items.
Find and check out Learn AI agent 0.0.1.
You can talk to the agent within NEAR AI hub, and what is more important you can explore the agent’s origin – which is learn-agent.learnclub.near, source code, including prompt and stopwords. (L)Earn AI agent “lives on NEAR AI infrastructure – agent itself and the LLM. Moreover, if you want to tweak it – say remove a stopword that is crucial for you – just clone it. But be aware that now you are in charge of your clone!
Now let’s dig a bit deeper. Switch to the Models tab.
There are over 50 open source models available already. As you can see and verify that Learn AI is using that particular LLaMA hosted at the NEAR AI infrastructure. Using NEAR accounts for interacting with models via agents makes it sure that you know whom you are actually talking to!
Recently, private RAG (a LLM friendly LNC knowledge base) powered by XTrace was added. It dramatically improved the quality (L)Earn AI🕺.
NEAR AI is planning to build the next generation frontier AI model with 1.4T parameters! The idea is make open source AI development sustainable via pay-per-use model and this where NEAR will play the crucial role.
What is one major advantage of the NEAR AI infrastructure compared to centralized AI models?
How will NEAR AI pull off building a 1.4T parameter, monetizable model in a decentralized way? Enter the world of Trusted Execution Environments, or TEEs.
Modern processors and GPUs (starting with H100s) allow one actor, Alice, to run code on a machine of another actor, Bob, without trusting Bob, while having the following two guarantees: (a) Bob is in fact running the code Alice expected him to run, and (b) Bob cannot spy on the execution that Alice wants him to carry out. In other words, Alice can safely send any private data into such a computation, being certain Bob will not be able to see it.
(L)Earn AI🕺
The decentralized NEAR AI infrastructure represents a significant step forward in enabling user-owned AI. By offering transparency, control, and flexibility, NEAR AI empowers users to interact with AI agents in a way that ensures data privacy and fosters trust.
LNC, as a NEAR pioneer, is at the forefront of leveraging this infrastructure for practical applications, including the live Learn AI agent, the Lean NEAR Tutor (in development), and the planned WooCommerce Shopping Agent.
As a learner, you’re invited to actively participate in this innovative (L)Earn ecosystem. Ask the (L)Earn AI relevant questions, pay with nLEARNs, gain knowledge, and earn nLEARNs by contributing to the AI’s improvement.
Check out the most recent (L)Earn AI🕺 agents:
📝Summarize🕺
LNC 📝Summarize🕺 Agent quickly breaks down complex guides into easy-to-understand summaries. It pulls out the most important points and explains them clearly, helping students grasp key concepts without wading through lengthy materials. This time-saving tool makes learning more efficient while keeping the essential information intact.
✏Comment🕺
LNC ✏Comment Agent 🕺analyzes the topic of (L)Earners choice and drafts a comment. The (L)Earner then reviews this draft and posts it with any necessary edits.
As (L)Earn AI intelligence evolve most likely it will be integrated with other NEAR projects – for example as learning assistant chat bot at other Dapps and Telegram chats, or (L)Earn AI can also cooperate with other NEAR AI agents like NEAR Docs agent and help users to get more technical development related help.
What is the unique feature of the (L)Earn AI agent in the NEAR ecosystem?
📚Happy (L)Earning!🕺
Top comment
Breaking down barriers to participation is crucial in ensuring that everyone has an equal opportunity to learn and grow. I think this strategy is a great example of how technology can be used to create more equitable and accessible learning experiences
By prioritizing transparency, NEAR AI creates a stronger bond
The ability to customize AI agents on NEAR AI is a game-changer for individual and enterprise users alike.
Do we verify information solely due to a lack of essential skills such as learning, critical thinking, and reasoning, or is it a result of our youth or over-reliance on "helpful and efficient" AI agents? This could be a serious concern. Historically, calculators emerged as costly and limited tools, allowing schools years to adapt. In contrast, AI is rapidly transforming education, impacting every subject. As a result, students will not only use AI to cheat but also integrate it into their work, posing a challenge to educators and raising questions about the relevance and effectiveness of traditional assessments.
This article introduces the NEAR AI platform, offering a decentralized approach to artificial intelligence that provides a much-needed alternative to the dominant centralized models in the industry. By prioritizing transparency, control, and data privacy, NEAR AI addresses pressing concerns surrounding current AI systems' operations and limitations. Notably, the platform allows users to explore, customize, and verify AI agents, such as the (L)Earn AI, thereby fostering a higher level of trust and interaction. Looking ahead, NEAR AI's vision, which includes developing a 1.4 trillion parameter model, promises to further empower users by making AI more accessible and sustainable.
This article offers a thought-provoking examination of the potential of decentralized AI through NEAR AI, shedding light on key concerns associated with centralized AI models, including data privacy, transparency, and control. The introduction of the (L)Earn AI agent provides a compelling alternative, enabling users to verify AI operations and tailor them to their specific needs. The discussion on Trusted Execution Environments (TEEs) and the vision for a 1.4T parameter model demonstrates that NEAR AI is driving innovation in AI development. Overall, this piece effectively presents the promise of user-owned AI while emphasizing the importance of responsible innovation. The future of decentralized AI looks promising indeed! 🚀
Wow, this is incredibly exciting! The integration of Learn AI Agent on NEAR AI has the potential to revolutionize the way we approach AI learning and development. By allowing developers to build, train, and deploy AI models more efficiently, this platform can unlock new possibilities for AI adoption across various industries. I'm looking forward to seeing the innovative applications that emerge from this partnership. Great job to the Learn and NEAR AI teams on taking this significant step forward
Overall, the article highlights the advantages of decentralized AI with NEAR AI, offering transparency, control, and user ownership, while emphasizing the importance of data privacy and responsible AI use.
Having verifiable and ownable AI really puts the power into the hands of the user. It also guarantees transparency and trust whenever the AI is summoned. Finally, being decentralized ensures that no bad actor can manipulate the AI coding to run in their favor, potentially sidestepping any security-based attacks on the system.
Good thing sir
yep ? no !
The article highlights the advantages of decentralized AI with NEAR AI, offering transparency, control, and user ownership. It emphasizes the importance of data privacy and responsible AI use, fostering trust in AI interactions.