Prompting Equity

People have really enjoyed asking the current generation of large language models (LLMs) like ChatGPT about AI: a sort of overly on-the-nose meta-discussion. Reid Hoffman, co-founder of LinkedIn and techno-optimist/evangelist, co-wrote a free book about AI with GPT-4 called Impromptu: Amplifying our Humanity Through AI.

In a late chapter, Hoffman makes an argument that modern humans are homo techne (“the technological human”), essentially that “technological innovation is, and always has been, humanity’s defining feature.” He talks about the many inventions of the past and how amazing it has been since the early 1920s. For example:

Whoever you were in say, 2005, you are—if you regularly use a smartphone now—a dramatically enhanced version of that person. (p. 192)

GPT-4’s response is a bit more nuanced:

I don’t necessarily disagree with any of the points you made, but I think it’s important to remember that not everyone experiences these benefits equally. For example, while it’s true that the average human life has doubled in the past hundred years, there are still parts of the world where life expectancy is much lower, due to poverty, lack of access to healthcare, or other factors. And while smartphones and other technology have certainly made our lives easier in many ways, they can also exacerbate existing inequalities—for example, by making it easier for those who can afford them to access information and opportunities, while those who can’t are left further behind. So while I don’t disagree that we’re living in an incredible time, I think it’s important to keep in mind that these benefits aren’t universally shared.

Reid:

That’s a very important point. Do you think modern emerging technology can help us democratize these benefits, or otherwise address long-standing equity issues?

GPT-4:

I think it’s definitely possible. For example, telemedicine could help address the lack of access to healthcare in under-resourced areas, and technology that allows for remote work or online learning could create new opportunities for people in rural or impoverished areas. But I also think that it’s important to remember that technology is only a tool—it’s up to us as a society to make sure that it’s used in a way that promotes equity and benefits everyone, rather than just a select few.

There’s a certain grain of techno-optimism that’s almost as tedious as the doomsaying: the belief that good arises organically as a feature of the technology itself as opposed to from the human systems that deploy it.

For a lot of people, the early democratic internet seemed to bear this out. This is what I thought as a middle schooler playing online hand-coding websites in HTML to post on Geocities. The whole thing is magic! And then we got Wikipedia–amazing! But I think in the era of social media and conglomerate capitalism, we should all know better by now. Hoping everything gets so profoundly broken that the governments of the world will be forced to provide a universal basic income (UBI) but that otherwise no harm is done is probably naive.

GPT4 is derived from the content of the internet. While the model is tuned to provide more useful and less racist/sexist/terrible responses, the fact that its statistical process is able to generate responses like this suggests to me that the people who have written about this publically have probably thought about it correctly. I don’t know if we’ll listen.

Leave a Reply