Here’s an argument that AI runs out of steam in the medium future, plateauing in ability gain

While this matters when considering the most extreme AI risks, we don’t need any AI capabilities past those we have today to have major disruption (good & bad) to work & education, etc.

In a lot of ways, this is an ideal use for AI. Since it is manipulating the data directly using Python, hallucinations are not a problem (though it rarely can misreport numbers when reporting out). And AI’s relentless problem-solving & clear explanations make it an easy coworker.