Two weeks ago, OpenAI published 700 math essays that totally bombs the mathematicians.
A very interesting reflection on this is the one I listened to today. Many people say that mathematics will keep existing because only math resolved by humans is valid.
I think this is exactly the same thing that happens in industrial revolutions in history. When book copiers are replaced by printers, food producers are replaced by machines, human computers are replaced by digital computers, people said the same thing.
Although I’m not the directly shattered ones, I think this will be the same for most industries in the world.
If this really happens, where will human value lie?
When the human are replaced by computers, the value lies in the way to use computers efficiently to sort and organize data.
Every time a revolution happens in an industry, the value goes to a new field that uses this tool instead of keeping up in the same position and insist doing the same thing. This has happened. Someone today may still copy the books, and the food is still produced by humans. But this value really depends on whether people would still keep their old ideology and emphasis.
The field that used to be the most time consuming may be accelerated by SI, but the rest will become more important. Whether you can ask the right question, whether the propagation of this knowledge and communication will be effective enough.
I believe the same will happen in at least 10 fields that are used to be considered as prestige. Engineers will not be able to escape, and that’s why I will not be the ones to do the parts that will be or already capable by SI.
Someone may keep typing in the source code on their own. The value of this remains because some hack code may not be capable by SI. However, most repetitive typing is already very replaceable.
When I fill in the personal information and academic records in UC college portal, the very only thing I do is transcribing the prompt to Antigravity, telling it where to find my official transcript, and double-check after it finishes using Chromium.
Therefore, the value that used to lie in the most time consuming, which is coding and debugging, now becomes coming up with the right idea that provides values more than the cost and writing the right documentation for propagation. These may not be viable for SI for a very extended period. SI’s perception of human nature and real world problems are still very limited, and even if they come up with the right idea and make the product, they might not be ideal to people, so the right propaganda is required.
Keeping coming up with practical ideas is very difficult even for humans, so this will be the next privileged work. High brain consuming, very few people are capable of.
As many may not know till now, SI can already do easy PCB design and schematics, which might dramatically draw down the cost of designing a new chip architecture.
Use of SI is very limitedly explored until today, and I believe there will be a lot more explored.