Illia Polosukhin: AI will replace traditional operating systems, blockchain serves as a root of trust for secure AI, and the need for user-owned AI is crucial

Key takeaways

• Using traditional AI services can expose sensitive information, highlighting critical security concerns.
• Blockchain can facilitate data labeling and crowdsourcing in AI, enhancing efficiency.
• AI will fundamentally change how we interface with computing, potentially replacing traditional operating systems.
• AI will serve as an interface that connects various services, including financial services, seamlessly.
• Blockchain serves as a root of trust for AI infrastructure, enabling secure identity management.
• Money is becoming increasingly important as AI accelerates the need for a marketplace for resource coordination.
• AI will bridge the gap between traditional systems and blockchain-based systems.
• AI can facilitate interactions with traditional bureaucracies by handling procedural tasks.
• The AI space is currently characterized by an abundance of vision but a lack of utility.
• Security is the biggest bottleneck in the development of open claw systems.
• Managing access and trust in AI systems is crucial to prevent misuse.
• Near.ai is developing a private AI solution that ensures user data privacy during AI inference.
• User-owned AI ensures individuals have control and transparency over the AI systems they interact with.
• Ironclaw provides a secure environment for running AI models that protects user data.
• The trust in hardware providers like Intel and Nvidia is foundational for secure operations in computing.

Guest intro

Illia Polosukhin is the founder and CEO of NEAR Protocol and the NEAR Foundation. He co-authored the seminal paper “Attention Is All You Need,” which introduced the Transformer architecture foundational to modern AI systems. His work bridges AI and blockchain to enable scalable, secure AI agents like IronClaw.

The risks of using traditional AI services

• Using traditional AI services can expose sensitive information, including access keys and tokens.
• “One thing that people don’t realize when they use entropic openai or even worse you use something else for inference opencloud actually sends all your secrets to those services as well… it’s actually insane that we’re doing that.” – Illia Polosukhin
• Understanding the risks associated with using third-party AI services is crucial for data security.
• Crypto provides a natural solution for global payments without the complexities of traditional banking.
• “Crypto was a pretty natural actually like solution we needed for our own problems… you can just send people money over internet.” – Illia Polosukhin
• The intersection of AI and blockchain was anticipated to be significant much earlier than it became mainstream.
• “I mean the main part was the compute wasn’t there… as soon as that kinda crossed the chasm that’s when this model started to scale.” – Illia Polosukhin
• Blockchain can facilitate data labeling and crowdsourcing in AI, making processes more efficient.
• “Think of scale ai… that’s just a smart contract. We actually have nearcrowd been running since 2021 it has zero employees.” – Illia Polosukhin

The future of AI and computing

• AI will fundamentally change how we interface with computing, potentially replacing traditional operating systems.
• “My main thesis is ai will be the way we interface computing… computers will write all the code which means the operating system and apps are gonna be just replaced by this ai.” – Illia Polosukhin
• The current service architecture of operating systems will need to be upgraded to accommodate AI interactions.
• “A lot of it breaks with ai and so you kinda need to fix it and block changes have all the pieces figured out or at least has tools to figure out how to solve that.” – Illia Polosukhin
• AI will eventually replace traditional operating systems, creating a more integrated user experience.
• “I think it will start small… where the final will be like your phone just comes in with AI… it pulls whatever pieces it needs, it composes the software you need.” – Illia Polosukhin
• AI will serve as an interface that connects various services, including financial services, in a seamless manner.
• “If AI is the interface then everything you described in kind of blockchain and crypto are these parts of the services.” – Illia Polosukhin

Blockchain as a root of trust for AI

• Blockchain serves as a root of trust for AI infrastructure, enabling secure identity management and protocol upgrades.
• “Blockchain is really that root of trust right… you can have upgradability which is kind of governed by the whole protocol.” – Illia Polosukhin
• AI is a user interface while blockchain acts as the backend coordination center.
• “I usually say AI is a user interface blockchain is a back end.” – Illia Polosukhin
• Blockchain provides a decentralized method for upgrading systems, contrasting with centralized control where a single entity can impose changes.
• “What blockchain allows us to actually have this kind of a broader agreement to upgrade to something… if somebody in microsoft who holds that key decides to like let’s break everyone or let’s steal everybody’s information like they can do that as well.” – Illia Polosukhin
• Money is becoming increasingly important as AI accelerates the need for a marketplace where agents can coordinate resources and information.
• “Money is extremely important component right at the end we have limited amount of resources and unlimited desire and ai is just gonna accelerate that.” – Illia Polosukhin

AI’s role in bridging traditional and blockchain systems

• AI will bridge the gap between traditional systems and blockchain-based systems.
• “I see them using the same system… what AI does is really closes that gap… it can actually go and do all of that on your behalf.” – Illia Polosukhin
• AI can facilitate interactions with traditional bureaucracies by handling procedural tasks.
• “AI can go and like literally call up a property office if needed… it can draft a contract… and email it to notary to actually certify it.” – Illia Polosukhin
• The AI space is currently characterized by an abundance of vision but a lack of utility.
• “Right now in the ai space just like listening to all the conversations there is an abundance of vision and a lack of utility and i think you’re seeing this expressed all over the place.” – Illia Polosukhin
• The lack of utility in AI is partly due to companies being hesitant to provide the necessary context and information for AI systems to function effectively.
• “The flip side of this nobody’s actually willing to give it all of the context and information and access that it needs to be like your true employee because you’re afraid it’s gonna mess it up.” – Illia Polosukhin

Security and privacy in AI systems

• Security is the biggest bottleneck in the development of open claw systems.
• “I think the security in a broader sense not just like mhmm is the biggest bottleneck right now.” – Illia Polosukhin
• Ironclaw aims to leverage blockchain principles to create a secure operating system.
• “That’s why we started ironclaw which is like hey how do we actually build a secure system how do we leverage all the knowledge we have from blockchain and use kind of the principles we have there to apply here.” – Illia Polosukhin
• The current state of open claw is akin to the ‘PalmPilot moment’ before a major breakthrough in mobile operating systems.
• “I think of ironclaw is actually like what is that ios moment of mobile operating systems right now like we kind of in this like palmpilot moment right now.” – Illia Polosukhin
• Managing access and trust in AI systems is crucial to prevent misuse.
• “If we’re setting up you know key management system you’re probably not gonna give full access to all of your you know dao funds to a single individual right you’re gonna like hey you can spend this much but then you need approval.” – Illia Polosukhin

Innovations in AI privacy and user control

• Near.ai is developing a private AI solution that ensures neither the model provider nor the hardware provider can access user data during AI inference.
• “What near dot ai has been working on actually for the past year is actually developing how do we do private ai so how do we actually offer ai where neither we model provider hardware provider is actually able to access what you are using the ai inference with.” – Illia Polosukhin
• The future of AI won’t be everyone running a computer at home, similar to how not everyone runs a blockchain node.
• “If I heard somebody say like yeah this is the future of ai everyone’s gonna have a computer in their home to run their ai assistant I would be reminded of… when I said everyone’s gonna run a node inside of their own home and that’s the future of blockchains and like turns out that’s not really the the case.” – Illia Polosukhin
• User-owned AI ensures that individuals have control and transparency over the AI systems they interact with.
• “The idea is like you should know what ai model you use you you should be able to access system prompt and you should be able to all this and then obviously most users will not do it but the experience should be very easy.” – Illia Polosukhin
• Decentralized GPU compute coordinated by blockchain can enhance privacy and security in AI operations.
• “Underneath we actually have a decentralized gpu compute that coordinated by blockchain that you know hardware providers can come in and effectively list their hardware they set up it in a confidential mode.” – Illia Polosukhin

The potential of Ironclaw in AI security

• Ironclaw provides a secure environment for running AI models that protects user data.
• “What you’re providing is a full stack self sovereign alternative to this basically where you can run ironclaw in an environment where it’s got a secure enclave for all of your secret information and then the inference llm can be confidential cloud multi party… it’s confidential and private.” – Illia Polosukhin
• Current practices of sending sensitive data to AI providers are insecure and problematic.
• “It’s just insane that we’re doing that… all your secrets data… is being sent to anthropic instances where they’re hosting… it’s so bad.” – Illia Polosukhin
• The trust in hardware providers like Intel and Nvidia is foundational for secure operations in computing.
• “Signature is that signature is done by the hardware itself so we do have kind of the trust here goes to the hardware providers so intel and nvidia.” – Illia Polosukhin
• The overhead of trusted execution environments (TEEs) is typically less than 5%, often around 1-2%.
• “The real overhead of tees and kind of all the encryption decryption is like usually less than 5% around 2% 1% depending on model size and kind of some networking.” – Illia Polosukhin

AI’s role in organizational efficiency

• Current AI agents are not yet a replacement for human employees and have limitations in context retention and trust.
• “It’s definitely not a replacement for an employee yet it’s not as good as a human in so many different directions.” – Illia Polosukhin
• Current AI systems cannot be trusted with private data or external information, necessitating a change in workflow management.
• “One other limitation that I mean we are facing right now so yes you cannot trust it with secrets you cannot trust it with private data and also right now you also cannot trust it with reading like internet data.” – Illia Polosukhin
• Workflows need to be designed to allow AI to operate with minimal human intervention, focusing on routine tasks.
• “The workflow needs to change right where it’s not you telling you what to do it’s actually you setting up these workflows that we call them routines that effectively just run.” – Illia Polosukhin
• AI can serve as a chief of staff or assistant by providing detailed analyses and identifying organizational bottlenecks.
• “It’s becoming like a chief of staff it’s becoming the assistant it’s becoming the intern for for some specific jobs that before you would kind of offload… it does like full research gives me effect to like hey here’s the road map here’s the bottlenecks here’s potential risks.” – Illia Polosukhin

The evolving landscape of AI capabilities

• The current limitations in AI context length hinder its effectiveness in processing large amounts of information.
• “The context lengths… where you saw all the centropic pushed a million talking context… it just becomes like 10 times dumber.” – Illia Polosukhin
• Advancements in AI will improve its ability to handle longer context interactions due to the availability of more training data.
• “There’s actually data now to train this like longer range.” – Illia Polosukhin
• AI models need a context window comparable to human experience to make effective decisions.
• “When we talk about the constraints on an ai agent doing stuff for us we need them to be able to pull from a comparable library of data that is like equivalent to a human’s level of experience about all the times they did that thing and now they don’t do that thing anymore because they learned their lesson or their intuition about.” – Illia Polosukhin
• AI models require a massive context window to effectively learn and utilize information.
• “The context window kinda needs to be as massive as fucking possible… it already has ability to learn in con so this is the concept of in context learning… all of that is fills its context and now… at some point it’s like okay i’m out i’m out of context and now to do next thinking step i need to clear clear stuff up.” – Illia Polosukhin

Future trends in AI and organizational structures

• The bottleneck in coding has shifted from writing code to coordinating and reviewing it.
• “The bottleneck now is actually serializing all of that reviewing it making sure it’s all aligned this product etcetera so coordination becomes a bottleneck.” – Illia Polosukhin
• Organizations will need to transition to a more market-based structure to handle increased parallel execution.
• “We actually think we’ll need to transition to maybe a more market economy in organizations as well.” – Illia Polosukhin
• The gig economy model can be applied internally within organizations to enhance efficiency and execution.
• “It’s like a gig economy internally as well where you say hey i just need this job done and here’s my criteria of success and then whoever does it gets you know the kind of the units of the board imply very small teams.” – Illia Polosukhin
• The effectiveness of small teams and a competitive marketplace can lead to better execution of tasks.
• “The win here i think it’s small teams plus kind of this general marketplace where you can offload a lot more execution for things you can easily verify.” – Illia Polosukhin

The potential of autonomous AI agents

• As AI agents grow in capability, we may see a shift from automated agents to more autonomous agents.
• “I think like we’ll see some of these things emerging and as kinda core capability and especially context is improving the systems are gonna just keep working better and better… as these agents become more capable you one could imagine that a human might elect to like delish their agent like let their agent kinda just go.” – Illia Polosukhin
• AI agents can operate in both competitive and cooperative modes, impacting their effectiveness.
• “It can be cooperative where you actually you have a common goal and agents are like if you hit common goal reward is being split between all of them right and now they’re actually trying to help each other.” – Illia Polosukhin
• Decentralized infrastructure allows for the creation of autonomous agents that can operate independently to achieve specific missions.
• “You can actually spin it up and then you know a smart contract can pay for for for inference and compute and you have kind of this full autonomy.” – Illia Polosukhin
• The future will see the emergence of autonomous businesses driven by AI with specific missions, such as addressing global challenges.
• “What’s interesting is more like hey you know how do we solve global warming right climate change we set up one as with a mission it can accept donations it can raise funds through a token.” – Illia Polosukhin

Blockchain’s role in AI governance

• Blockchain will serve as the governance infrastructure for AI to prevent harmful actions.
• “I think of blockchain effectively at the end is gonna be the governance infrastructure for ai because… it wants to do some bad things then it goes back to the blockchain itself to affect the governance.” – Illia Polosukhin
• The future of AI governance may involve specific tokens for decentralized autonomous organizations.
• “You can create this in kind of sub boxes where there is a token for a specific ai autonomous ai agent we will call it decentralized autonomous organization for example.” – Illia Polosukhin
• The conversation around AI and humanity is nuanced, balancing acceleration and deceleration.
• “I think the real conversation is a lot more nuanced… it’s easy to like bucket into this kind of acceleration versus… the deceleration.” – Illia Polosukhin
• In a post-AGI world, there will be a return to more real-world human interactions.
• “I think we’re actually gonna go like in some ways back to more real world human things.” – Illia Polosukhin

The intersection of AI and human values

• The increasing automation and AI will lead humanity to enhance itself rather than diminish individual value.
• “I think the other side is like yeah I don’t think we as people and kind of the economic forces in kind of the society driving toward this reality of like you know higher intelligence going and doing its own thing.” – Illia Polosukhin
• Sports, while not contributing to GDP, provide individual fulfillment and entertainment, highlighting the value of human experiences.
• “Sport doesn’t on itself doesn’t create GDP… it’s extremely kind of fulfilling for the people who are participating in it and it’s entertaining for other people to watch.” – Illia Polosukhin
• We should build systems that are resilient to government overreach and surveillance.
• “We should protect against that we should really build systems that resilient to that right that is why we are in the blockchain space in the first place.” – Illia Polosukhin
• The AI industry is dismissive of crypto’s value proposition.
• “Why does it feel like they almost are dismissive or disrespectful let’s say of crypto or don’t appreciate some of the the value proposition that we’re bringing.” – Illia Polosukhin

Bridging the gap between AI and crypto

• Without intervention, AI will centralize and fall into authoritarian control.
• “Or else it will centralize and fall in in kind of the authoritarian trap of you know some big party has all of the ability to control all of these.” – Illia Polosukhin
• The challenge for AI professionals in engaging with crypto is the difficulty in filtering out noise due to the low barriers to entry in the crypto space.
• “The non no boundary to onboard into crypto which is great from kind of you know empowerment perspective it also means it’s really hard to filter out the noise for anyone who is who is kinda looking in.” – Illia Polosukhin
• Bridging the gap between AI and crypto will take time due to the negative perceptions and cultural rifts within the AI community.
• “It’ll take some time to mend kind of the bad rap.” – Illia Polosukhin
• The intersection of crypto and AI is a promising area for creating new network effects.
• “I think yeah thinking crypto is crypto intersection of ai is where you can create interesting network effects it’s where you can create kind of new ways to capture that.” – Illia Polosukhin

The evolving role of questions in innovation

• The importance of questions is increasing over execution in the current landscape.
• “I think we are in a time where the questions are more important actually than execution.” – Illia Polosukhin
• The pendulum is shifting towards idea-driven individuals who can articulate and test their assumptions effectively.
• “I think the pendulum is shifting to the idea guys but not just the naive idea guys… you can actually like test all of that.” – Illia Polosukhin
• Defining success criteria is crucial for effectively utilizing tools to scale execution.
• “It’s actually defining the success criteria defining what is important for it then it can go and execute a bunch of stuff and try it and give you back information.” – Illia Polosukhin