Generative AI is the buzzword these days. A KPMG survey in the USA expects generative AI to have a tremendous impact on business. However, most business are not yet ready to adopt it immediately.
A Goldman Sachs report questions the use of generative AI in business. Big Tech is set to spend $1trillion plus on AI capital cost. There are doubts about the pay offs of this capital outlay.
The current scenario is a ‘picks and shovels’ one, and AI is yet to become a game changer. Despite its potential, it is to be seen how far it delivers.
Daron Acemoglu had an interview with GS. He is a celebrated writer and an MIT professor. He expects that in the next 10 years, the productivity and growth from generative AI is going to be limited. He estimates only 25 per cent of AI exposed tasks will be cost effective in next 10 years. The implication is that AI will impact less than 5 per cent of all tasks. There will not be tremendous improvements in AI models, and they will not be become less costly in near future.
The effect on the GDP roughly translates into a 0.9 per cent impact over the decade according to Acemoglu.
Human inventions are celebrated, and AI is one such invention. Still the technology is not yet ready for prime time. AI facilitates automation. However, the issue whether such automation is needed.
To earn sufficient return, AI must solve complex problems. AI does not have this capability at present. AI investment in the next few years could be $1trillion. The investment consists of investment on data centers, utilities and applications. What problems would it solve? Can we apply AI to low wage jobs?
AI could be compared to early days of internet. Amazon started selling books more economically than Barness & Noble. Brick and mortar structures were costlier. Today the Web 2.0 provides solutions such as Uber. Whether AI will be able to match this? AI technology is expensive and could be sustained only when it could be applied to complex problems.
AI today sustains on GPUs provided by Nvidia. High cost is in-built in it.