Four months ago, when GPT-3.5 was still known as black technology, and the big language model was still a cutting-edge technology that only a few people understood, I introduced an AI assistant.
With the threshold of getting started without learning, the point-and-click interaction method, and the straightforward solution to the "rigid needs" of workplace formalism for the majority of migrant workers, this AI assistant can be described as a great success. Although it may seem like a simple H5, the number of active users exceeds that of many flashy apps.
In the past few months, the AI assistant has undergone several minor repairs, nothing more than adding more dialogue scenarios, and still maintaining the original intention of the simple, lightweight and efficient product.
With the birth of more and more AI tools, I originally thought that AI assistants had completed their stage mission and should gradually withdraw from the historical stage.
However, in fact, the more people I contacted, I found that on the one hand, there are still many people who do not know what these big language models can and cannot do, and on the other hand, there are many inexplicable courses that are popular in the market...
That is to say, half of 2023 has passed, and even GPT-4 has been figured out in structure. There is no doubt that today's AI is not a panacea and cannot replace human jobs; What we have to do is to build on our strengths and avoid our weaknesses and use them for the most efficient things.
So, we did another upgrade, still the same principle, no need to learn, click out-of-the-box, and direct connection to GPT-4.
The interface and functions include four modules: dialogue, creation, simulation and painting.
The new dialogue function calls the 3.5 model by default, which is very responsive, and new registered users have a free usage quota, which can increase the number of free times by inviting new users.
At the same time, it is possible to switch directly to the GPT-4 model. Don't play virtual, real 4. The response speed is slightly slower, but the quality of the answer will be much higher, more logical, presumably everyone has heard more or less, take the writing of articles, 4 will indeed be more literary than 3.5, more logical and complete.
Dialogue is just an appetizer, and in the creation, we preset many, many dialogue scene functions.
How many dialogue scenarios are there? If you roughly count, it's about 505.
Yes, it is to set the identity of the AI assistant with a preset prompt word, click to enter to start the conversation, and you don't need to write any additional prompt words to use.
Writing daily reports, weekly reports, articles, summaries, translations, video scripts are all basic operations, have long been commonplace, we have also added some interesting scenarios, such as name analysis, SEO suggestions, illegal word detection, duplicate checking, speech script, salted fish copywriting... Too much. Whatever prompt he has, just click on it:
Simulation is also a new feature, more or less inspired by character.ai, and such a tool is prepared for everyone.
23 characters are preset, such as interviewer, fitness trainer, swimming coach, cosplay Monkey King and more. It's both a form of entertainment and a bit of a helping one, depending entirely on how you use it.
This part depends on your feedback, and some interesting character cosplay may be added in the future.
There is also an AI painting function that supports Chinese conversations, about 30 seconds a picture. Although not as professional as MJ or SD, occasionally a small picture can still be entertaining.
Finally, the entrance to the number of white conversations is here, and the rules are very simple, so please check it yourself
By the way, the way to get into the AI assistant is also very simple:
You can follow me and enter "ai" (case-insensitive) in the message to get the entry address.
Without magic or complex prompts, this wave of updates directly flattens everyone's cognitive information gap. What else to hesitate, hurry up and try it.
Happy use!