How to Replace Scientists with AIs - The Crazy Guys
It only takes two steps.
Have you ever thought about how scientists work? I have. Scientists solve problems with some patterns. Some people call the ‘styles’. I am sure that different scientists have different styles. On the other hand, there must be limited categories of ‘styles’.
With that in mind, the first step clearly should be the data collecting phase. You set up a company to help Ph.D. track their research process. To get data that can be analyzed, the company should provide a tool to track their research process. And the tool should also be useful to the researchers.
Here is a simple example of such tools. I developed a note-taking system that is like flowcharts or UML. However, it’s optimized for research.
Each of the node should be linked to a lab note about the specific trial.
Of course the visualization has to be optimized and done automatically.
The next step is to collect all these data and find out how researchers work out problems. Combined with the actual outcome of the researcher, we define successful researchers. Train a machine to behave like these researcher and replace every scientists.
I told this to Josh and he replied that we can train a bunch of scientists with different research styles.
Yeah that would work. We can build a community of virtual scisntists to work out all kinds of research problems.