2023-08-12 AI writes research notes daily Sentences generated from joint editing of and by the system
本文
Diverse perspectives] are required to improve knowledge productivity. AI can provide a broader perspective.
At this point, focusing on systems that produce value in a stable manner is narrow-minded. It is beneficial to actively collect uncertainty of the type discussed in “antivulnerability,” which is a set of limited downside risk and blue-ceiling upside risk. Such a system does not create value in the short term, but it is “more haste, less speed.
The current usage of AI is to provide AI with “verbalized” information as a prompt. However, in this way, it is not possible to share tacit knowledge (knowledge management terminology), which has not yet been verbalized, between human and AI. Rather than having the language first, the human and the AI should first conduct “collaboration” to share the experience. For this purpose, a cycle that includes collaboration, such as the SECI Model, rather than the PDCA cycle, would be beneficial.
感想
- Indeed yes, [[this itself is a real example...]]. - Intellectual production support using ChatGPT tends to increase information, but this is not reasonable because the "amount of information that can be targeted at one time" by humans is limited - The KJ method transforms information from a wide field of view into a form that can be processed by humans by repeatedly compressing the information into table-tops. - This process is KJ-legal in that it does not involve sticky notes or two-dimensional placement, but "makes it processable by humans through repeated compression. - The major realization of the original manuscript was that "planning" in "sharing tacit knowledge before planning" was abstracted into "verbalization" and led to a verbal prompt to ChatGPT - The concept of an AI system that puts humans in a cycle is not uncommon. [[Human in the loop]]. - However, I think it is often put in the role of C in the PDCA cycle. - Interesting to put it in the S of the SECI model, i.e., to make AI and humans equal and engaged in the same task. - This will make humans neither "verbalizers before the AI" nor "judges of outcomes after the AI". - Not sure of the connection between the second and third paragraphs? - We believe that it is worthwhile in the long run to collaborate with the AI, to let the AI and yourself live in the same place as the same kind of being. - But that value is not visible in the short term. The system cannot be designed to ensure that value is obtained at the push of a button. - Trying to produce [[stable value]] goes against [[making AI equal to humans]] in the first place. It is trying to [[instrumentalize AI]].previous version(version -1)
[[🤖2023-08-12 07:08]]The content of the article was passed to ChatGPT GPT4 as a memo to be turned into an article: [chatlog](https://chat.openai.com/share/43ef9c6e-f98e-4a16-aa4c-6db52a1b3cd7) - After seeing this, finally reworked the opening sentence - > Combining these perspectives, a new paradigm of knowledge production for the next generation will be developed. - After looking at this and thinking about it, "planning" in "sharing tacit knowledge before planning" abstracted into "verbalization" and led to "verbal prompts for ChatGPT" Based on these notes, I will find new relationships and draft a blog entry in Japanese.new relationship
- PERSONALISM AND WHOLE KNOWLEDGE PRODUCTION: Incorporating the perspectives of not only humans, but also other life forms and AI will improve knowledge productivity.
- Anti-Fragility and Knowledge Production: The idea of anti-fragility, which benefits from uncertainty, acts as a new approach to knowledge production.
- Information organization methods and the KJ method: Through reflection on the KJ method, improvements in information organization and analysis methods can lead to more efficient knowledge production.
- Collaboration and Knowledge Sharing: The relationship between collaborative editing projects, the SECI model, and the PDCA cycle indicates that collaboration and tacit knowledge sharing are important for knowledge production.
- Anxiety Tolerance and Knowledge Production: Anxiety tolerance, learned from the saying, “Slow and steady is as good as fast,” is a factor that improves knowledge productivity.
Draft blog entry in Japanese
. A wide variety of perspectives are required to improve the productivity of knowledge. Learning from person theory, it becomes clear that a broader perspective is needed, not only of humans but also of other life forms and AI.
In addition, the idea of anti-fragility is proposed as a new approach to exploiting and profiting from uncertainty. This will also work in conjunction with the way information is organized and analyzed, paving the way for more effective knowledge productivity through, for example, KJ method reflections.
The importance of collaboration must not be forgotten. Exploring in depth the relationship between collaborative editing projects, the SECI model, and the PDCA cycle underscores the need to share tacit knowledge before creating a plan.
Finally, we also consider the importance of anxiety tolerance in knowledge production from the perspective that “skillful delay is better than poor speed.” In this line of thinking, the spirit of haste is the key to improving the quality and quantity of knowledge.
The perfect combination of these elements for knowledge productivity will be a force for creating new value. Combined, these perspectives will develop a new paradigm for the next generation of knowledge production.
version -2
What outputs while I sleep - The process of "do some work, go to sleep, and then wake up and see what the AI has developed based on that work" could be useful. - The concept of [[person theory]] was introduced at this time.from 🤖2023-08-12 07:08 Thinking about parsimony, we may need to take a broader view that includes not only humans, but also other life forms and AI to improve the productivity of knowledge. Person theory offers a new perspective for improving knowledge productivity.
The concept of antifragility also provides an important perspective for improving knowledge productivity. The idea of profiting from uncertainty can be a new approach to improving knowledge productivity.
Reflection on the KJ method points to an important issue in the way information is organized and analyzed. This indicates the need to improve the way information is organized and analyzed in order to increase knowledge productivity.
Also, considering the relationship between re-titling and the SECI model and the PDCA cycle in collaborative editing projects, it is clear that in order to improve knowledge productivity, it is necessary to understand the importance of collaboration and the importance of sharing tacit knowledge prior to planning.
Having a “skillful and slow is as good as poor” perspective requires a certain amount of anxiety tolerance. This indicates that a certain level of anxiety tolerance is necessary to improve knowledge productivity.
version -3
nishio.icons] looked at the raw output that nishio.icons had put out, selected the ones that looked vaguely good, cut out the superfluous words, and collected the ones that looked vaguely related. The following is a collection of those that seem to be related.
- The latter five groups are only collected and left alone because I could not find a parent element that could encompass them, although I collected them and looked at them.
- Anxiety tolerance is necessary for “skill is better than speed.” I wrote about it on another page because it was interesting.
from 🤖2023-08-12 03:01 To generate new perspectives and ideas, it is necessary to go beyond existing frameworks
- The capitalist-labor rivalry was established at a time when the means of increasing productivity was the purchase of production equipment by capital. Today, however, knowledge has become increasingly important, and this antagonistic structure has become obsolete. This indicates the difficulty of updating the frame of thinking.
- The KJ method of reflection points out the problem of information overflow and messiness. This is an important issue to consider when thinking about how to organize and analyze information. Another problem noted is that it does not generate new ideas, using only the parts that match the existing framework of ideas.
Uncomfortable metaphors have the potential to generate new perspectives and ideas by expanding them before denying them. This can be a useful technique for creative thinking and idea generation.
Anti-vulnerability is the state of being able to benefit from the positive aspects of uncertainty without suffering serious harm from the negative aspects of uncertainty. This is an important perspective in risk management and strategy development.
The concept of derailment = association = new combined discovery is considered an important process for generating new perspectives and ideas. This is considered an important process for increasing freedom of thought and generating new perspectives and ideas.
- When will you change the front page of the KJ Act?
- Exciting” because “something new” is created that was not anticipated.
- It is considered an important process for generating new perspectives and ideas.
- Re-titling in collaborative editing projects
- Communication Issues
- Challenges when people with different perspectives and understandings work together
- The Importance of Collaboration
- Also related to SECI model
- Regarding the relationship between the SECI model and the PDCA cycle, S (collaboration) is necessary before PDCA
- Share tacit knowledge before creating a plan to create a plan that makes sense.
- Regarding anxiety tolerance, this shows the connection between human psychology and behavior. This is also related to the interpretation of “skillful and slow is better than poor” and the re-titling of the project in the joint editing project.
- Having a “skillful or slow is better than poor” perspective requires a certain level of anxiety tolerance. People with low anxiety tolerance tend to be “zero-risk oriented” and may have a tendency to pursue short-term success.
version -4
At this point, [[antivulnerability]] , [[Derailment = association = new combined discovery]] , [[When to change the front page of the KJ method (creation process)]] , [[ KJ Method Study Session @ Cybozu]], [SECI Model and PDCA CycleAccording to the experimental logs, he tripled the information loaded on the prompt at this time.
-
I had somehow set the upper limit for loading existing sentences to 2000, but after thinking about it, I decided to load up to 6000 since the GPT4 context can go up to 8K.
-
You’ll always be given an input of about 6,000, and you’ll be given an output of 2,000 tokens, so you’ll compress it by a factor of 3. - AI writes research notes daily
from 🤖2023-08-12 02:30 The idea that abstraction is not necessarily a good thing indicates the importance of keeping hooks for recalling concrete things and being able to associate with a wide variety of things. This is also related to the associative meaning of words to connect with and why the KJ method from the survey is so difficult. In addition, the quote from the Analects of Confucius, “To err and not amend, this is called a mistake,” indicates the teaching that making mistakes is human nature in itself and unavoidable, but recognizing the mistake and not amending it is the true mistake. I feel this is an important lesson to be learned in the research and design process as well.
The idea of recruiting directors from within the company is based on the point that the position of director usually serves as an expert in management skills and an elder in the internal order, but this is not the case for Cybozu’s openly recruited directors. While there are concerns about delays in decision-making and bias caused by having more directors, Cybozu’s culture of “try it and stop if a problem becomes apparent” suggests that if something goes wrong, you can just stop. This may be related to the suggestion of backing up the map itself easily. There is a suggestion that it would be good if the map itself could be easily duplicated and backed up to cover situations where you want to make a major configuration change but are afraid to do so. This was essentially impossible when we were working on paper, but digitization has made it possible. Backups would have the effect of alleviating the fear of making changes.
The idea about anti-fragility is to take safeguards in one area (= robust against negative black swans) and take lots of limited risks in another area to get big gains from big wins (= leaving positive black swans). This is believed to achieve anti-fragility. Taleb also states that anti-fragility is about having options. Options are what make us anti-fragile. The idea that “options allow us to benefit from the positive side of uncertainty without being seriously harmed by the negative side of uncertainty” is an important perspective in risk management.
The concept of derailment = association = new combined discovery is considered an important process for generating new perspectives and ideas. This is considered an important process for increasing freedom of thought and generating new perspectives and ideas.
The idea of when to change the front cover of the KJ method (the creation process) is presented as “exciting” because it creates “something new” that was not expected. This is considered an important process for generating new perspectives and ideas.
The relationship between the SECI model and the PDCA cycle is mentioned, with the assertion S-PDCA that S (collaboration) is necessary before PDCA. This is because a plan is made by formalizing tacit knowledge that has not been verbalized and combining it, and if the tacit knowledge is not collaborated and shared in advance, it will become an “Unconvincing plan”.
version -5
[[🤖2023-08-12 01:16]] I won't go back any further because the content is almost irrelevant to the final results of this project.This page is auto-translated from /nishio/AIとの共同化 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.