1 Comment

I think we should be cognizant of teh fact that AI in its various forms has been available for many decades. t has become increasingly useful as each stage matures. The current hot (hyped) technology is LLMs which due to its easy user interface (text prompts) has caught the imagination. While some evamgelists have said it increases their productivity, it remains to be seen if it will really do that or not. I expect it will for some areas of work.

However, as we have seen with decades of software, that doesn't translate into measured productivity gains for the economy, and certainly not in wages. The big concern is that currently LLMs are extremely energy intensive, both in training and deployment. IDK is that is going to be acceptable.

I expect to wait for the hype cycle to reach the "plateau of productivity" before we know the place of LLMs and very similar technology in the economy, and whether it will show useful productivity gains to boost GDP, or be more domestic and be primarily used at play, like spreadsheets, wordprocessors, browsing the WWW, and "photoshopping" images.

Expand full comment