The recently departed from OpenAI, chief engineer and genius (he wrote
this
fantastic paper at Columbia university at 17 and got the price for the best economic Thesis...) has written a number of essays analyzing the computacional power density, the orders of magnitude of change in the speed of model training and development and the geopolitical element of "winner takes all" in the US-China war to reach AGI... and it is scary, how fast and how easily his predictions can become a reality. Alignment is therefore a MUST in order to have a mild effort to have some control over the superintelligence(s) that Leopold sees apearing in 10-20 years max. But today everyone has a different definition of alignment, and companies are still more concerned by profit and governments by being the first at creating "the ultimate weapon" so we are far from secure ... Taking this thinking to a positive avenue, if we survive, we'd have a world of overabundance, deflation, full employment, based on Nuclear Fusion energy and powered by Quantum computers that help us develope the superintelligence that spans from AI... But to take advantage of this change, we need to change our mentality of growth so that we are not bound ONLY by profitability... as the
Conscious Capitalism
group advocates