When I worked at IBM many decades ago (in a npn chip making plant as a co-op), they wanted to use some complex artificial intellengence concept, and
we played with it a lot, but found very little utility. However, when we used very simple programs, tools, and macros to automate routine tasks so
people could avoid lots of data entry, analysis and boring report generation, it was a huge success.
I found the same results in chemistry. I saw several attempts to build complex chemistry automation systems that would try to do complex reactions
or multi-step syntheses. Most failed spectactularly, some just clogged and jammed a lot. The more complex, the harder/more expensive to build,
maintain, and use. But when we took small tasks and bought machines to do them, like weigh vials, inject samples, collect fractions, etc, they were
very useful and helped productivity. So while I like the idea of a genius computer that can be asked to design a drug, find world peace, or write
poetry in some ways, I don't think it is what I need. What I need are tools to let me avoid typing the same data in 5 places, write lab reports,
order my chemicals (based on the amount of each I need to do a certain synthesis), and predict the weather accurate enough to make real plans on.
My problem is that often even if you have good data and great predictions, bad policies, managment, and government regulations seem to require things
to be done a certain way, even if there is a better solution. It does me no good to have a computer tell me how to make a compound if I cannot order
the chemicals to make it, have regulations forbidding having them, or my management tells me to do it the wrong way.
|
![[*]](images/xpblue/default_icon.gif){kind=icon}
