A common refrain among the most vocal advocates for AI is that critics simply don’t understand the technology. Less technically-minded people (and, frequently, the scientists, scholars, and engineers who disagree) are simply operating under an ignorant set of outdated assumptions. The technology is so complicated that even those working on it, we're told, don’t understand it!
While LLMs are an incredibly sophisticated technology, I don’t believe it’s fundamentally impossible for us to understand how they work. More to the point, I think the idea that things like LLMs are just too complicated for ordinary people to understand is, at least in part, a PR myth designed to disempower would-be critics.
I’ll explain my thinking on this point in more depth in another post, but suffice it to say that, in the same way that you don’t need to memorize a neighborhood-level map of redlining in Chicago to understand that housing plays a role in reinforcing structural racism, I don’t think you need to learn matrix calculus in order to criticize modern AI technology.
As for me, the more I’ve come to understand about LLMs, the less I think the hype is justified.
To be clear – and you’ll probably hear this refrain again – I’m not a software engineer or computer scientist. I’m a speculative fiction writer with an academic background in religious studies and a longstanding interest in technology, especially those places where the remarkable and sophisticated abstractions undergirding modern technology intersect with our social, political, and experiential realities.
So I’m no expert, but I’ve spent a lot of time thinking and learning about this stuff over the course of the last few years, and I want to try to make this process a little more public in the hope it’s useful to people. I can’t promise I’ll always be right, and I’m always open to feedback. But when it comes to the vast AI hype machine, a lot of the self-proclaimed experts clearly don’t know what they’re talking about, either – especially when the technology intersects with the questions we study in the humanities. I don’t know if knowledge is power, but I’ve found it empowering both to learn more about the technology and to recognize the value and relevance of the different kinds of knowledge I already have, and I hope my posts on this blog empower you in the same way.
- Theodora