Diary 2024-03-30
Concepts I'd like to organize but haven't had time to get around to.
nishio I'd like to organize but haven't had time to get around to it, concept, I'll do it on my phone while on the train. AI for domestic demand and AI for external demand
Tradeoff between ease of use and cleverness of AI
nishio Domestic and foreign demand is a metaphor Needs inside and outside the company
Daikin had a big enough need in-house, and there weren't enough people to create something to meet the need, so they created a training institute.
Microsoft has 220,000 people in its own company and had been doing DX before AI to improve the efficiency of its operations, which is also a big need in the company.
nishio How to meet external demand No security of distribution without clear product packaging, poor comparability of products, severe information asymmetry, and markets not functioning properly
nishio Example of collaboration between PFN and Fanuc, PFN had highly specialized personnel within the company, but there was no internal demand that could be met by them. Like the Nordic countries, if domestic demand is insufficient, we need to look for external demand, and that's where we connected with the demand.
Same thing with MS now sending engineers to companies that are in demand.
nishio Demand needs to be uncovered itself, not clearly verbalized demand first. Because technical expertise is needed to convert a "sense of the problem" that is not clearly verbalized into a technically "solvable problem.
On the other hand, this "sense of challenge" is also not clearly verbalized, so it is not marketable and cannot be transferred separately from the field.
nishio The need to combine "resources that are difficult to market because they are tightly coupled to the field" with "resources that are difficult to market because they are tightly coupled to the brains of experts", which is what the " The decision is being made that this can only be achieved by "sending experts to the field". nishio Consider the relationship between ODA and nishio on the trade-off between ease of use and cleverness. The hardest to use is the smartest.
It's the human policy constraints that create the difficulty of use.
nishio There are players with and without policy constraints, and the reason why the more constrained means are used is because they are better. Players with policy constraints try to optimize within the constraints, and the inferior performance is chosen, this is not a bad thing, and sometimes the inferior performance is good enough nishio paid-for barriers, external cloud service usage policy barriers, border barriers, risk aversion barriers... nishio When some policy constraint results in inferior performance using an inferior system, the listener may not understand that the "inferior performance is due to a policy constraint. This is an unfortunate mismatch. nishio The technical side recognizes that policy constraints on the part of the listener are lowering performance, and the demand side thinks, "Technically, you're still not performing that well. performance" and the demand side thinks "the technology is still not performing that well". nishio I think we need to visualize what happens when there are no policy constraints by avoiding policy constraints all the time to achieve high performance, and then I think it would be better to aim at lifting policy constraints based on that. I had claude put it together.
---
This page is auto-translated from /nishio/日記2024-03-30 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.