The idea is good, but the ideal is beautiful, and in the next sentence, the microphone is handed to you, and you say it yourself.
In theory, this idea can be realized, but it is also limited to theory, because to put into practical application, many conditions are required, and the cost of many conditions may be higher than hiring a lawyer, here is not a single cost, but an average cost.
First of all, the first condition we have is to feed data, and it is massive data. We have noticed news reports, often said that Tesla in order to achieve unmanned driving, I don't know how much real human driving data collected, to achieve unmanned driving, is to machine according to human thinking mode to think, deal with complex situations, by feeding the machine massive data, let the machine continue to learn, so as to evolve the human neural network, and then continue to train, replace humans.
To replace lawyers, machines need to be fed a lot of litigation data. These data types can be much more complex than road surface processing, because the output of road surface processing is nothing more than accelerator, brake, and steering. But for litigation, including criminal, civil, administrative litigation; In the case of criminal proceedings, it also includes both procedural and substantive matters; In terms of substance, there are many different crimes, which may be beyond the statute of limitations, jurisdictional objections, not guilty defenses, voluntary surrender determinations, and so on.
What lawyer would give this litigation data to a machine? It can only be acquired by capital at a high price, and if the cost of acquiring data is too high, the comments are distributed to the user, and the cost is unknown. But Tesla's charge for opening a seat heating and other functions cannot be said to be complaining, and it can be regarded as criticized.
Secondly, the second condition we must have is that users must learn to use AI. Needless to say, there will be smart people who will start carpentry immediately: "Since it is called AI, it is fully intelligent", which is not a rough word. You can't see it, the Internet is overwhelmingly teaching you how to correctly use AI to write advertisements for articles, novels, drawings, etc., why? Because although it is called AI, it is also like people, it needs to eat and pull, only eating without pulling is Pixiu, only pulling and not eating does not conform to the law of conservation of energy, quantum mechanics, and the law of increasing entropy.
For the corresponding computer, even if it is intelligent AI, it is impossible to directly output the results you want with a look from Ke Daxia. At this time, how to enter conditions has also become a science. At that time, such as the type of case, procedure, substantive defense, and so on.
Again, the third condition we have is how to read the papers? Criminal case files are state secrets and cannot be leaked, and if you bring an intelligent AI with a network function to read the files, you are afraid that you will be rejected by the procuratorate and the court. If you can't read the papers, you can't know yourself and the other party, how to achieve an effective defense?
Here are just a few nonsense points, in fact, there are many conditions, all of which require high equipment costs, labor costs, time costs, and some possible conditions may not even be achieved at all.