The lay person's skepticism of AI's usefulness

A significant part of my work is helping organizations use data and AI efficiently and effectively. But how about lay-people? In my experience, the challenge there is different. While organizations are convinced they should do more AI, and the work there is more focused on the how, regular individuals we are still in the why phase.

I would understand the feeling if we are still in early 2023, where the tools were still slow, hallucinating, without access to up-to-date information, limited by short context windows and other issues. But now?

Example

Recently, I had a discussion with my father on politics and history. We were discussing Marx, and he shared a question he had for a long time, that he wasn’t bothered to research:

"In Das Kapital, is there any mention by Marx on whether an entrepreneur (capitalist) should pay themselves any salary?”

This is a significant question, since it would show an interesting nuance. Das Kapital is more than 2000 pages long. A simple keyword search would also not help - we are looking for meaning between the lines. So for a project like this, even if you are familiar with the author’s work, you can easily spend a whole weekend.

Naturally, I used AI to find the answer:

“In Das Kapital, Karl Marx doesn't provide a clear or consistent example of the entrepreneur (or capitalist) paying themselves a salary in the modern sense. His focus is on the capitalist mode of production, and he generally treats the entrepreneur or capitalist as someone who appropriates surplus value generated by workers, not as someone earning a wage.”

So far so good, but this is where it becomes interesting:

“When he discusses the functioning capitalist, he sometimes acknowledges they perform real labor (like coordinating production), and that they might pay themselves a salary as a deduction from profit. But he sees this as a disguised form of profit, not true "wages" in a labor-exchange sense.”

This nuance was very interesting to both of us, so we asked where in the text we can learn more, and the AI obliged and even gave us a link to the original text in digital form:

“Look to Volume III, Part 4, Chapter 23: "Interest and Profit of Enterprise".

And lo and behold, that was correct, and we could learn much more from the chapter:

“[…] from that portion of profit which remains, after interest is deducted, as profit of enterprise in the salary of management of those branches of business whose size, etc., permits of a sufficient division of labor to justify a special salary for a manager.”

Conclusion

This is just one of the many examples of how I get more relevant, well-referenced, up-to-date information and interpretation on topics that would take me hours to research.

I realized the issue is that so few of us still actively involved in knowledge management, and even using a search engine productively is a rare skill, and these tools are so much more than search engines. Perhaps a good avenue to increase adoption among lay-people is to focus on why they should be doing more knowledge work, instead of just impressing with the technology?