By the time you understand a new technology, it's too late to steer it
A 1980 diagnosis from a British researcher still describes every debate about regulating something new, before the something new has decided what it wants to be.
In 1980, a researcher named David Collingridge described a trap that has since been named after him, and it shows up every time a society tries to get ahead of a new technology instead of just reacting to it.
When a technology is new, it's cheap and easy to redirect, restrict, or kill outright — but almost nobody can yet predict what it will actually do once it's everywhere, so there's no clear case for intervening. By the time the consequences are obvious enough to justify action, the technology has wired itself into factories, habits, cities, supply chains, and other technologies that now depend on it. Stopping or redirecting it has become enormously expensive, maybe impossible. Collingridge summarized the bind himself: when change is easy, you can't yet see the need for it; once the need is obvious, change has become expensive and slow.
Cars, leaded gasoline, social media, and industrial agriculture all followed some version of this arc: early enough to redirect cheaply, nobody had the information to justify it; late enough to have the information, redirecting was a fight against entrenched infrastructure and interests.
There's no clean escape from the dilemma, only tradeoffs. Move early on thin evidence and you'll sometimes strangle something harmless. Wait for certainty and you'll get it only after the technology no longer needs your permission to keep going.