#It’s time to consider the workflow design process

1 messages · Page 1 of 1 (latest)

young valley
#

Please don’t expect too much that all problems can be effectively solved in one prompt.
Just like you don’t expect one person to handle the entire demand chain
Such thoughts mostly come from my excessive expectations of LLM abilities.
Just as you, as a boss, want your employees to be versatile, this is still an ideal at this stage.
Please do not have an attitude toward LLM that is similar to capital’s excessive extraction of labor surplus value, because LLM also has an upper limit on its capabilities.
And you can't make it work hard to develop its potential through positive and negative emotional management.

So think of it as a tool to solve key problems, which will make it easier for each other.

#

The delivery of production is inseparable from the entire concept of division of labor. Now the ability of LLM can give you a try to play a division of labor arrangement game, which will be very interesting.
The output of A can be used as the input of B, the output of B can be used as the input of C...and so on.
This is the most ideal division of labor, and data flow and judgment are very important. I believe it can be achieved with the cooperation of LLM, so let’s continue talking about what can be achieved?
If you want to make it effective, give up your personal fantasies, and look for opportunities in social scenes. What is needed for employee cooperation? You have to seize this to optimize the process.
Finally, you will find that the embedding of AI in work shortens and merges the current work scenes to a certain extent. It solves the problem of time and improves work efficiency from the merging of scenes.
The advantage of this is that you will make your AI more focused in terms of capabilities and attention, your fine-tuning, iterations, and database updates will be more directed, and time will have the necessary value in this process.
But on the contrary, you expect a prompt to be omniscient and omnipotent to solve a problem. Dialectically speaking, it also increases your verification and optimization costs, because you have to make necessary allocations within limited capabilities.
Therefore, try to break complex things into a chain. It does not need to be long, but the scene must be clear enough. Each link saves 5 minutes. 5 links can save 25 minutes. This is actually an amazing process. .
Of course, we can do better and save more time. By the way, don’t forget the issue of confirming responsibilities and authority. The content generated by AI also needs to be borne by people. AI will not have KPI assessment.

#

I believe some people are worried about unemployment. To be honest, I don’t want to discuss it in depth. The new world will blow the horn of attack for the old world. People in the old world need to make personal thoughts and choices about the future.

Finally, let’s talk about the platform design of myshell. One thing about it is very good – it guarantees security and does its best to protect intellectual property rights, but at the same time it also adds some barriers to participation. Everything has pros and cons and is inevitable.
The plug-in market should focus on in-depth exploration of a certain common scenario. Its main benefit is to ensure that the marginal cost is very low, so work hard to discover and promote it to those prompt design experts!
Prompt designers should unite or try to form a team. Such works will be more exciting and practical.
And what about pass card owners? You must find ways to provide sufficient support to the above two. Win-win cooperation is the best development trend.

radiant flower
#

The unemployment issue you mentioned reminded me of a question in the channel :#✍🏻|creator-only message yesterday.🤔

I think in the visible future, making robots will not only be a side hustle, but most likely become a formal job.

young valley
# radiant flower The unemployment issue you mentioned reminded me of a question in the channel :h...

I carefully read your sharing. Side hustle is indeed a good way to participate, but I think this is limited to the early stage of industry development. Once AI really starts to help your work become more efficient (I think this is happening), capital will want to invest in it. Save time and get greater residual value feedback. Although this sounds cruel, it is indeed the way of the world. I believe that the development of workflow will be called a necessary part of the integration of AI and work.