Does Ethical AI Use Exist?

Karen Hao’s Empire of AI never asks, “Is it possible to ethically use GenAI tools?” But the way she laid out the many real-world “externalities” (to borrow a phrase from my b-school days) from OpenAI’s obsession with scale has left me pondering for weeks.

It’s a familiar question from other parts of life. Wrestling with the global effects of consumption—from coffee to running shoes to electronics to social media—is a topic that has been hashed and re-hashed for centuries. The speed with which OpenAI created a new category of demand, however, and the far-reaching effects of that effort ripped the question open for me to re-examine.

I’ve come to live with the fact that I need shoes, after all; but can I say the same about GenAI tools? Is my analyzing data faster worth the brutal mental health toll on data annotators in Venezuela and Kenya? Are tools like DeepMind’s AlphaFold worth that price?

And If employers expect and require use of GenAI tools, do we actually have a choice to opt out?

For now, I’m in a space where I still feel like I do need to engage with these tools. But Hao’s book has left me with new dimensions to think about how I use them, which ones I use, and the standards I should judge them by.

I’m still trying to understand the tools that will help me do that (if you have any recommendations, please share), but knowing what you don’t know is the first step.

If you have some free time coming up over the holidays and want to better understand how the world has changed, and the philosophy that drove those changes, over the last few years, buy a copy and check it out yourself: https://karendhao.com/