Personally I consider some of the most interesting things happen
with a technology a few decades after it is revealed. Often this is
because the time wasn't correct for it, then at a later time, a time
came along that was correct, which happened to be that time at
time. Often, however,
it is simply because a patent was in place,
nobody moved forward on it out of fear, until long after all of this
cooled down. Eg, the invention of the transistor occurred almost
exactly as soon as some prior patents expired.
Now, I know patents protect the hard work of an individual and
reward the individual or their owner by giving a period of time
which to capitalise upon their hard work in discovering the thing.
However, this tiny funnel in which one person can get a bit more
wealthy than the rest of them for quite some time isn't really
overall for humanity (it's only useful for that one or few people
the patent). The real work begins after the expiry, where the
contribution of a great many other minds on the topic can produce
diversity, efficiency and overall improvement.
Instead of protecting all the hard work in inventing stuff, what I'm
suggesting is use robots. Using AI to say 'no need to put in any hard
work in the discovery, I'll discover it all for you, make apparent
once was occluded, you can just get on with using it'. I think this
accelerate human development considerably, as a patent is about
third to a quarter of a lifetime, in general. Think how much
ahead we'd be otherwise.
AI should soon be able to, within a specific domain, exhaust
configurations then deduce and document resulting
thus 'getting there first'. This obviously won't work for everything.
I've been looking at the many and varied configurations of
transistors, some of which are named after the one who got there
first. The Widlar Current Source; the Wilson Current Mirror
after a ball); the Blackmer cell; etc.
These are quite creative configurations, each quite simple in
imagine there may be some further useful configurations, as yet
hitherto undiscovered or undisclosed to us because we've not
'invented' them yet (ie, realised they were possible, all along).
also because too many modern designers tend to just bolt
ICs without regard to what's in them and how they work at the
fundamental level). Imagine an AI that could run a wide variance
possible configurations under a fanout of conditions, and take
of how useful each one is, for whichever uses it can imagine. This
the hard part, not the rapid run through of permutations -
what 'useful' might mean. However, imagine we'd arrived at a
definition of useful, and it used that. I suggest we'd end up with at
least a few more undiscovered useful ways of connecting
If it can do it to a defined domain such as electronics, there must
a lot of similar defined domains that an AI could permutate
and rapidly exhaust all reachable outcomes of 'usefulness'. Then
can just get on and enjoy life, using the stuff instead of trying to
find where most of it is hidden first, then when someone finds it
behind the curtain or down the back of the sofa, it remains
untouchable for most of the rest of your working life.