How AI Captured All of Twitter's Attention
The first wave of AI hype hit Twitter in 2022, with the first announcement of ChatGPT by OpenAI. It was the first full-fledged chatbot that used open sources of information and gave a full logical response to users. Many saw a gold mine in this and began to launch their projects, labeled as artificial intelligence. Naturally, such projects had no future, since they were essentially selling the same ChatGPT, only in a different wrapper.
The second peak of popularity came in 2024. One of the first AI agents was AIXBT - He quickly began to gain popularity due to communication with the audience, calls on various coins, analytics and daily posts on the social network Twitter.
AIXBT currently has over 450,000 Twitter followers. How can a single AI agent handle so many incoming requests?
What is the mechanism of operation of AI agents?
1. The code that controls the LLM (Large Language Model) logic is run.
2. A smart contract launched in a decentralized environment controls the agent’s behavior based on the actions of the holders.
3. Voted for changes in the behavior model or for the sale of this or that asset, the author and holders receive income from what the agent earned.
That is, the entire economy and mechanism of work is based on the holders of the agent token and the creator who developed the agent.
What is the main weakness of AI agents?
1. Weak economy:
We have already talked about the mechanism of AI agents in Web3, and herein lies the main weakness - an insufficiently developed economy. Many promising projects related to AI died at birth because they were unprepared: they did not receive due publicity, the idea did not have time to be realized (narrowly focused area), an insufficiently developed economic model.
All this kills most AI agents, which is why potentially interesting technologies do not reach the masses. AIXBT exists only due to the huge number of holders, although at its core there is no real foundation for future development.
Another example, Infera.
The project has good subscriptions, an interesting concept and good potential for future development, but at the moment token fell by more than 80%.
Perhaps with the arrival of large players with serious AI agents performing complex analytical functions, such projects will see the light of day, but at the moment memes are too dominant.
2. Weak results:
On Twitter, you can often see posts from various influencers, in which they describe in detail what parameters to set in order to earn on AI agents, show how this or that token took off, and the AI agent was able to predict it. But what is the reality?
An interesting investigation was conducted - original sourceThis study took over 220 AIXBT signals from January 3 to January 10, with return time windows of 5 minutes, 10 minutes, 6 hours, 24 hours after the tweet was posted.
The following results were obtained: the proportion of correct signals during the first hour after the tweet was published was 60% for 44 signals, but after 24 hours the indicator dropped to 50%.
It can be concluded that by following the AIXBT signals, you flipped a coin and with a 50/50 probability: either you made money or you lost.
3. Lack of utility and actual benefit:
At the moment, 99% AI agents on the market are dummies, copying each other and dying within a few days of existence. The best results are shown by platforms that exist for the creation of such memes. It is quite appropriate to compare with pump.fun, where a meme is created for the sake of a meme.
What is the future of the AI agent narrative?
At the moment, the trend for AI memes is gradually declining, although new major players are emerging on the market - for example, DeepSeek and Qwen2.5 (Alibaba), which can spark a new flame in the AI agent trend and inspire the creation of real technologies in the Web3 world that can change the understanding of the cryptocurrency market.
You can follow the latest and most relevant news about AI in our telegram channel Hare AI.
