AI dreamers and dream-busters are locking horns amid the market frenzy. Optimists place their faith on extrapolation, taking recent innovations as a starting point, while pessimists in the logic of history, with disdain-dripping jibes at "this time is different" messages, laced with the tales of cryptos.

Lessons drawn from cherry-picked historic episodes, assuming all indisputable, is a favorite method of skeptics unwilling to invest time in understanding an innovation. That most innovations heralded as pathbreaking fall severely short is the percentage they rely on.

The jumbles, called GenAI foundation models, are unprecedented. Their existence alone proves pessimists' logic wrong, even if conclusions could still be right. In a world full of writers listing how AI will change whatever space they focus on in the next few quarters to decades, in the least, risks of utter disappointments are real.

Optimists feel that with more processing power and training, AI-models have to be on an exponential path of progress. As discussed previously, this writer believes in potential of follow-on innovation waves in healthcare or autonomous vehicles, to name two, dwarfing the achievements so far.

Extrapolation is a perilous venture, given our incomprehension of how these models work. They produce extraordinary results. We can hypothesize the meaning of what we observe. But, nothing guarantees more spectacular results in quick time merely with the expansion of training datasets by a factor of ten or thousand. These models are nascent; they need substantial improvement before deployment in tasks more critical than document summary.

Nonetheless, the scales tip in favor of the optimists, albeit for a different reason. The circularities caused by the focus on AI's potential will fuel the momentum for years, with momentous events in between because of the efforts. Almost every major corporation globally is investing extraordinary resources in harnessing its potential. The magnitude of investments in AI could break all records, not just in absolute figures but also as a percentage of global GDP in a few years.

A viral report from a reputed consulting house last week charted over 85% of AI benefits accruing to the non-tech sectors. Previous posts by this writer explained how the lack of major IP barriers could cause a worldwide race of foundation models in the coming quarters, notwithstanding the current shortage of the compute.

Periods of stagnation without significant innovation developments are inevitable, and they will coincide with an increase in skepticism in the volatile financial markets. Yet, with so much attention of the world's brightest and the mightiest shifting to the potential of GenAI models, it may pay to stay hopeful. Just that in the grounds trodden by the giants and in technology waves so diffused and ill-understood, one may want to invest more in spaces that recognize these characteristics too.