Some of you may remember a newsletter that you sent a month ago, saying how a startup in China has released a new agent of artificial intelligence getting polite reviews.
After that, the Monica Emerg Company gave me early access to Manus AI with a lot of free credits to play with artificial intelligence agent.
Now that I got a direct experience, I can confirm, it really raises the game on the artificial intelligence front.
Its ability to determine the review list needed to achieve a goal to end and perform one task after another to get there to breathe.
So, the great task I chose to perform with Manus AI is to design a high -level game where the player lives throughout Indian history by trying to build cities and defend them.
Manus had clear instructions – we were getting to know the storms to reach the perfect gameplay, but then a viable game symbol only.
Instead, I tried to achieve the “goal” anyway in one implementation – and we ended up with this intended page of the game that I published on its own, instead of getting MVP from the way of playing.
If you think, I can only ask that refinement and return to what we were trying to do, uh, bad luck, we are all outside the credits.
Therefore, I ended up smuggling all these luxurious free credits that Monica gave me, so that we can learn this valuable lesson. We need a red button on artificial intelligence.
Artificial intelligence agents are all anger. I am very optimistic about them as a person who builds my multi -organization agents.
However, we need to recognize the negative aspect.
Artificial intelligence agents, and even those who enjoy the highest line, are not yet ready to perform complex tasks independently.
This should be on the label.
Because it is not possible to expect users to detonate $ 100 (or even 1000) of dollars before realizing this.
This is not just a case with Manus AI, though. This is a similar problem I face from time to time with the index-the transition to artificial intelligence of coding.
Sometimes, it offers unnecessary ingredients or even new libraries, completely different from what you currently use, in your code.
Take your eyes from what the agent does for one second, and you may end up with an epic dimensional.
Now, in order to be fair, the indicator has a “refusal of all” button. But this only works if you immediately review the code.
Often times, the problem may be noticed after doing more work on the code, then the retreat of Git remains the only option.
This does not discourage you from the experience of artificial intelligence customers (they are really the future), but just something I need to say loudly, especially since I contributed to the noise around them.
Artificial intelligence agents are currently an applicable option for a comprehensive symbol and publishing things like intended pages, classic games and in -depth research on any specific topic.
I just want this button to my office, when pressing, just stops any task implemented by artificial intelligence at one time, and everything back back.
This post is an excerpt from the twelfth edition of the artificial enhanced newsletter.
