What happens when, faced with a breakthrough in a transformative technology, we decide to pour all of our imaginable resources into the front-runner? I’m sure we could all think of some contemporary cases, but a historical example might be more informative.
Let’s take heavier-than-air flying machines shortly before 1900. Though we’ve seen a great deal of revisionism around this (much of it coming from management-consultant TED Talks), the most advanced technology in the field was very probably being developed by Samuel Pierpont Langley in the area of steam-powered aircraft.
https://en.wikipedia.org/wiki/Samuel_Langley#Aviation_work
His first success came on May 6, 1896, when his Number 5 unpiloted model weighing 25 pounds (11 kg) made two flights – 2,300 ft (700 m) and 3,300 ft (1,000 m) – after a catapult launch from a boat on the Potomac River.[11][12] The distance was ten times longer than any previous experiment with a heavier-than-air flying machine,[13] demonstrating that stability and sufficient lift could be achieved in such craft.
On November 11 that year his Number 6 model flew more than 5,000 feet (1,500 m).
Despite these early successes, Langley himself saw that internal combustion was the future of flight and by 1901 had commissioned what would turn out to be the revolutionary Manly-Balzer engine for his first attempt at manned flight.
Nonetheless, in 1896, steam-powered aircraft was the state of the art. What if we had put everything into developing this technology and building an industry around it, maxing out financial and research resources and effectively crowding out work in competing and adjacent fields?
Eventually, internal combustion would have won out (the advantages of the technology were overwhelming), but we would probably have ended up with a lost generation of progress not just in aviation but in all sorts of related areas. It would have been a huge waste of resources with tremendous opportunity costs.
It also would have done the exact opposite of what it was intended to do.
Right now, with respect to AI—and certainly with respect to natural language processing—large language models are the clear front-runner, and we are in the process of spending unprecedented sums trying to get them to the next level. This decision has been driven largely by hype and bubble mania, with developers being given both almost unlimited resources and permission to break whatever laws get in their way, ranging from intellectual property to environmental regulations.
It seems like the burden of proof should be on the boosters to argue that we aren’t betting much of the world’s economy on yet another steam-powered airplane.
No comments:
Post a Comment