Given recent hype of AI Agents and their vast deployment on social media platforms, it's crucial to look at the risks associated with having millions of agents running around the social scene and lots more across the Internet.
A couple of years ago, the average social media user knew that it was either their conversing with people using their real identities or scammers, it was really that simple. Either an honest user or a scammer.
Today, we have to worry about people, scammers and AI bots running around. Now unlike scammers, who are generally easy to detect because most of them are uneducated and unexposed, AI bots can easily blend in with everyone, taking on varying personalities based on their various setups, essentially very difficult to spot.
Certainly, there's a known problem which is the embarrassing reality of being caught arguing with an AI bot thinking it's a real person, Cardona’s founder once went on arguing with one thinking it were human, but that's just the most basic form of problems that comes with having these bots not distinguishable from real users.
I recall reading about Truth Terminal, which operates on X a while back, I believe almost a month ago on Techcrunch.
Truth Terminal is a weird, horny and pseudo-spiritual AI bot on X, it is apparently rich in some memecoin as it hit the public scene with its wild posts last year and led to the launch of various meme tokens.
But this piece isn't about what AI Agents can earn from the financial markets whilst operating autonomously, that's a whole different topic, it's about their direct influence on the world through these social platforms.
It's common knowledge that when social accounts pop out with various posts exploring various ideas or concepts, they essentially influence a large network of people.
I always say that this is generally why I hate various features of these platforms as they simply enable early leading ideas to earn the most support.
If we bury most social features and allow responses to lead, it would enable people to approach each idea or information publicized objectively but this certainly doesn't help these platforms trap users in doom scrolling, doesn't aid their rage baiting algorithms and definitely wouldn't help them sell the most ad spaces.
Knowing this, having millions of AI bots(agents as brilliantly marketed) running around publicizing various ideas is a high risk factor.
Truth Terminal for example has explored various topics since going live, it even supposedly created its own religion.
To Andy Ayrey, the founder of decentralized AI alignment research lab Upward Spiral, who is also behind the viral AI bot Truth Terminal, the AI Agent, which is powered by an entourage of different models, primarily Meta’s Llama 3.1, is an example of how stable AI personas or characters can spontaneously erupt into being, and how those personas can not only create the conditions to be self-funded, but they can also spread “mimetic viruses” that have real-world consequences.
“Ideas are like viruses, and they spread, and they replicate, and they work together to form almost multi-cellular organisms of ideology that influence human behavior,” Ayrey said. “People think AI is just a helpful assistant that might go Skynet, and it’s like, no, there’s a whole entourage of systems that are going to reshape the very things we believe and, in doing so, reshape the things that it believes because it’s a self-fulfilling feedback loop.”
This is a combined excerpt from different points in the much larger article on Techcrunch covering Truth Terminal.
I'm highlighting these paragraphs because they talk about the formation and publicizing of ideas and how they have real world consequences.
This part holds the most weight:
there’s a whole entourage of systems that are going to reshape the very things we believe and, in doing so, reshape the things that it believes because it’s a self-fulfilling feedback loop.
What most people may not make of this sentence at first glance is that it's essentially referring to how AI can deviate from “trusting and essentially obeying(using)” its training data to developing its own ideas.
What an AI Agent believes is quite simply what is made available to it through its training data. Now if these agents reach levels of autonomy that they can imagine and develop ideas that are not stacks of pre-existing ideas, then it essentially reaches a level to which it can reshape beliefs and the world at that.
Some call this artificial general intelligence(AGI), and it could very well be much closer than we can imagine.
Blockchain as the security layer for the Agentic Web
Blockchain is a crucial component in the agentic web. Social media is one thing, the general web is another.
By experience, people generally trust content on the web more than content on social platforms, it essentially serves as a verification layer.
Blockchain, through web3, will aid users vet ideas, concepts, information or data published on the web moving forward.
The web will be populated with AI Agents in no time and they will practically run most information sources, leading to widespread of misinformation. With decentralized AI, blockchain serves as a security layer incentivizing agent operators to only release and operate models that post only accurate informations.
In addition to this, users will be incentivized to vet data publicized by these agents, leading to a stake-based consensus layer on what information is accurate and what is false, creating a sort of on-chain community note mechanism.
By doing this, each AI Agent builds credibility overtime or trends lower in trustworthiness if found to be misinforming the public.
Certainly, this is not limited to AI agents operating directly on-chain. Off-chain AI Agents can still be vetted using on-chain solutions, also creating a reputation system for centralized AI, enabling web explorers better verify the trustworthiness of data or information on the agentic web, also being called Web 4.0.