I'm 80 and I wasted 25 years of my life. Don't make my mistake. - Howard Marks

He says his view changed after his son, Andrew, who works with AI daily, told him that “so much has happened” that the memo needed a full rewrite. Marks highlights AI’s ability to contextualize information about the user and communicate strengths/weaknesses in a way he finds exceptional.

Marks contrasts past innovations as productivity tools with AI’s autonomy—being able to assign a job without specifying the method. He also emphasizes AI’s unpredictability: he doesn’t think anyone knows the future shape well enough to forecast it confidently.

He argues AI will expose (“defrock”) people whose claimed investing skill is overstated, similar to how indexation revealed persistent active underperformance. He’s uncertain whether AI’s capability list is limited or effectively unlimited, which he frames as central to whether it can fully replace human investors.

Marks says he believes experienced investors still add value via judgment calls like avoiding “bad people,” where instincts and subtle cues matter, and by handling situations with no historical precedent to train on. He adds that some people are better at reasoning about probability distributions for future events.

Describing the period around Lehman’s collapse, Marks says they had no data or prior experience for “the end of the world,” only supposition, yet still invested because not investing would mean failing their job if the world didn’t melt down. He stresses they were not confident and that probabilistic thinkers act with trepidation.