ICOs, yield farming, NFTs, the metaverse, Ordinals and inscriptions, DeFi, DePIN, RWA, etc.
Various trends and concepts have gained vast attention over the years; some are justified, and others represent overblown hype.
The next one will most likely stick around and gain significant traction: AI agents.
These programs can gather data, connect to their environment and use the information to carry out independent tasks to achieve pre-determined objectives.
A popular example includes certain virtual assistants/chatbots used as companies’ first point of contact, especially corporations that handle numerous customer service requests.
Driverless vehicles, autonomous drones, and spam filters are other common applications of this software.
AWS provides a helpful outline of these agents’ key principles, benefits and challenges. Nvidia offers this post about AI agents in customer service.
Discussion and interest in AI (notably artificial general intelligence/AGI) have skyrocketed since the advent of ChatGPT in November 2022. Note that this is not an AI agent. Rather, it’s an AI model (i.e., a large language model in this case) generating content based on human commands.
How does it tie in with crypto and blockchain technology?
AI agents offer vast productivity improvements, particularly when conducting research and analysis by investors and traders. These sophisticated tools can also manage decentralised oracles, APIs, Web3 wallets, and other purposes in crypto and blockchain tech.
One of their biggest drawcards is being able to research, process, and utilise vast swathes of data 24/7/365. This abundance of relevant and helpful information will likely give these traders an even bigger lead over individuals with limited resources.
The recent explosion of AI-themed meme coins — many of which run on Solana — adds to the general fervour (at least in the short term) surrounding AI agents.
However, unlike the countless meme coins that come and go, these autonomous agents are not a fad.
These agents will become mainstream as technology advances, particularly as they become more sophisticated, reliable, and affordable (mid-range to high-end autonomous software currently costs hundreds of thousands of dollars to develop).
While trading bots have increased in popularity, they differ from AI agents. Unlike bots that follow specific rules for certain outcomes (deterministic models), the latter are much more flexible and constantly learn new information to make better decisions, using probabilistic models (ones that account for randomness and deal with likelihood).
I recommend this piece from CoinGecko to learn more about AI agents and how they differ from bots. It provides a deep dive into this emerging technology.
Related crypto projects and assets
One of crypto’s earliest and most renowned AI agents is Truth Terminal, which led to the creation of the Goatseus Maximus (GOAT), one of the plethora of meme coins released on Pump.fun, a Solana-based launchpad.
GOAT stands out as a bridge between AI agents and meme coins, where the enthusiasm behind crypto memes could accelerate the popularity of these agents.
Moreover, most tokens with any reference to AI will probably pump with the rest of the market, most likely at a faster rate.
In fact, during this period, I’d say the more nonsensical the name and project’s idea, the more likely it is to increase in price.
Here are a few protocols and tokens to watch in this space/
Fetch.AI ($FET*)
One of Fetch.AI’s many use cases allows people (including beginners) to create intelligent agents with the uAgents Python library.
Others can be set up for business cases, including:
— virtual assistants for shopping
— cleaning services
— table bookings
— message authenticity
— sending tokens
— asynchronous loops
*The token symbol on all exchanges and DEXes will use the ASI ticker symbol very shortly as part of the Artificial Superintelligent Alliance token merger.
AI Agent Layer ($AIFUN) is an emerging small-cap that could
According to its website, “each AI agent on its platform is tokenized and paired with the ecosystem’s native token ($AIFUN).”
This token is used as the main currency for purchasing agents. As more people purchase these using the protocol, the price of $AIFUN also benefits.
Furthermore, $AIFUN liquidity pools get a boost every time someone buys an agent through this channel.
The token is also used for staking to secure the network.
Thanks to Crypto Banter (Ran Neuner) and Ivan on Tech for bringing this to our attention.
Check out the related documentation to learn more about this, which also covers tokenomics and its roadmap.
Virtuals Protocol ($VIRTUAL) is a portal for creating and co-owning AI agents. The app divides them into two broad categories: Sentient and Prototype agents.
GAME (Generative Autonomous Multimodal Entities) ‘s platform and engine power Project Westworld, a simulation in Roblox. Learn more about this integration in this Substack post.
I alluded to this asset and project in the following article.
Dasha ($VVAIFU)
Described as “The pump.fun for autonomous AI Agents on Solana”, I believe it will continue its solid run over the past month (+3,400%) as this trend continues to pick up steam.
I anticipate some capital shifting from meme coins to these AI-agent-themed tokens for at least the remainder of this bull cycle, especially when degens are involved.
Remember to take profits along the way and convert gains into ETH, BTC, and SOL. The latter (and RAY) will continue benefiting from the Solana network activity and Raydium, its leading DEX.
Concerns regarding AI agents
Beyond the excitement about the huge productivity benefits AI agents can offer those using their services, these models are still in their infancy and require extensive high-quality and relevant training data to perform well.
This will be imperative when the stakes are high, and the margin for error should be as low as possible.
However, these tools are not foolproof and still require human oversight to ensure these systems output factual information, at least for years to come.
This will be imperative when the stakes are high, and the margin for error should be as low as possible.
A major issue with this AI software is data bias. Societally, it risks exacerbating prejudices against individuals or demographic groups.
The following example covers this phenomenon in recruitment.
“When AI systems learn from biased data, they can perpetuate or even amplify existing prejudices. For example, an AI recruiting tool might favor male candidates if trained on data from a male-dominated industry.”
SmythOS
Tying this into crypto, there’s a risk of algorithmic bias, where these autonomous systems might prioritise and cherry-pick certain narratives to create a filter bubble, in turn overestimating a project/token’s potential.
As this asset class is rife with pump-and-dump schemes coupled with misinformation, it’s important to do your due diligence before assigning capital to a coin or token.
There’s another important element to the use of this autonomous software, which is:
Who is legally responsible if these agents provide false information to a trader or investment firm, causing massive financial losses and reputational damage?
If a client uses this automated software by decentralised protocols, imposing (full) liability on a single entity is complicated.
There could be ramifications for co-owners of these AI agents, but I question how easily this can be enforced and, more importantly, whether they’d be at fault. This is a whole topic of discussion.
It’s akin to companies or governments not being able to hold Ethereum liable, nor holding Uniswap accountable, for bogus projects that run on these protocols and for accusing the latter of allowing “unregistered securities” on its DEX.
Additional thoughts
While these agents will provide a productivity boost to multiple sectors, we must proceed with caution.
These AI tools are a double-edged sword: While they present huge growth opportunities for investors and entrepreneurs, things can go horribly wrong if these agents get hacked, disseminate wrong or misleading information, or are used for other nefarious purposes, such as corporate and state espionage.
There will also be a significant demand for AI agents in robots for multiple purposes, such as providing medical assistance, data collection, companionship, private transport, etc.
Japan appears to be leading the way and embracing these devices figuratively and literally. I recognise these offer benefits, but it would be concerning if humans rely on them as a substitute for human interaction. At that point, it starts getting trippy.
How nuanced can this autonomous software be in assisting people as an affordable and effective alternative to conventional counselling or helping them deal with chronic loneliness? Once again, this is an article in itself and outside the realm of crypto and blockchain tech.
In conclusion, many AI-agent crypto protocols and tokens will stick around and make a big difference in various sectors. As this is a nascent aspect of crypto and blockchain tech (which is still maturing), it’s too early to tell which ones will make it.
Watch this space.
by
at Freepik. AI-generated content (unspecified base model).
P.S. I haven’t gone into as much detail about this subject matter versus other articles as the prices are rapidly rising. I expect a market-wide boost in the coming days on the back of upbeat CPI data, meaning we’ll likely see a rate cut for the final FOMC meeting of 2024.
Disclaimers
• N.B. None of this is financial advice; I am not a financial advisor. You are ultimately responsible for crypto investments, let alone in any asset class.
• I don’t have a computer science background and have never worked in this field. I am writing this from a layperson’s perspective as a crypto enthusiast.
• The opinions expressed within this piece are my own and might not reflect those behind any news outlet, person, organisation, or otherwise listed here.
• Please do sufficient research (DYOR) before investing in any crypto assets, staking, NFTs or other products affiliated with this space.
Featured image by Phonlamaistudio at Freepik