It is a fair question to wonder how global high-performance computing capacity might change if interests that want slower or little AI data center growth get their way. Right now, about 72 percent to 75 percent of global high-performance AI compute and supercomputer capacity, according to data tracked by Epoch AI and analyzed in reports by institutions like the Federal Reserve.
It is an equally-important question whether words about “reining in AI” will result in corresponding action, if the outcomes are negative. The adage “live by the sword, die by the sword” is relevant here.
At the same time, artificial intelligence and data center-related investments have driven a major portion of U.S. gross domestic product growth, accounting for roughly 30 percent to over 44 percent.
Also, AI investment has represented as much as 75 percent to 90 percent of capital spending in recent years.
The issue is whether politicians are willing to gamble that the industry can be throttled or scaled back without losing the positive economic benefits AI has been providing to the economy. As the old adage warns, in retail politics “it’s the economy, stupid.”
And even if it has been very difficult to measure AI impact on productivity, for example, new evidence is starting to emerge of quantifiable gains, even if aggregate output and employment levels are affected by lots of other inputs.
One expects a certain amount of political posturing by politicians. But one also expects those same politicians to make note of any corresponding policies that could negatively affect the economy. Nobody wants to be holding the “we tanked the economy” bag when the next election happens.
The point is that “AI winter” is not likely to happen, at least not because of draconian government policies. It’s one matter to say what voters want to hear at a moment in time. It is quite another matter to follow through on policies that actually make the economy worse. Voters will punish politicians just as they rewarded them.
No comments:
Post a Comment