On AI and reports

The current AI systems seem rather good in writing reports and analysis. So, we can get analysis and reports on interesting topics cheaply, rapidly and with little effort. Will we be better off now? Or very soon?

Let’s think what are the reasons for having a report on or an analysis of something. They are not valuable in themselves, but because we can understand some phenomena better, and together with our values, we can make decisions, we can change the state of the world. A report which does not help in making a reasoned, value based decision is worth nothing.

Now, when we get an analysis for free and without much effort, if any, and can turn the analysis into a report, can we make better decisions, faster? Are the decisions more informed?

In order to make a decision, we need to understand the report. What if the AI is too clever, or the user of AI too lazy? Then, there is no understanding, at least no new understanding. How can the user of the AI make a decision if they do not understand the reasoning in the report?

I think automating reports and analysis is commendable, but only if the users of the AI is willing to make enough intellectual work to understand that the AI has told them. Just having an AI to produce a report or an analysis and presenting it to some audience is rather pointless.

This is why I think AI will not make intellectual work obsolete. On the contrary, AI makes intellectual work more valuable, and thus also intellectuals!

(Picture of my desk in Boston back in 2005 before AI made studying easy)