AI for Activism
I’m not on the side of the AI-haters. Yes, I see the danger. And I see the positive potential for activists.
In 2019, OpenAI announced it had created GPT, an artificial intelligence that could generate text on any topic. Fearing that this AI would be used to influence politics, the start-up deemed their invention too dangerous to release.
Last year, OpenAI reversed course.
Today, anyone can access OpenAI’s text generation artificial intelligence. Moreover, OpenAI has made a user friendly version (ChatGPT) which has already passed business, law and medical exams—and terrified many whose livelihood depends on reading and writing text. The growth of ChatGPT has been unprecedented: with “over 100 million monthly active users in January, just two months after launch, [ChatGPT is] the fastest-growing consumer application in history.”
A wild success for OpenAI, and a new chapter for social change. As one scientist put it, OpenAI has released “a pocket nuclear bomb without restrictions into an unprepared society.”
OpenAI continues to acknowledge the potential harms of GPT. Underlining their concern, the company has published an analysis of how their AI might be used in disinformation campaigns and potential mitigations.
Officially, OpenAI has banned GPT from being used to “influence politics.” However, what makes the technology both alluring and dangerous is that it is not a simple matter to ban the AI from influencing politics. The technology doesn’t work that way. Instead, it is gullible. Tell GPT that you are writing fiction and it’ll oftentimes happily disobey OpenAI’s restrictions. And so while OpenAI won’t condone political uses of their technology, there is not much that they can do about it—as long as those uses do not require OpenAI’s approval.
I’m not on the side of the AI-haters. Yes, I see the danger. And I see the positive potential. Sometimes I go so far as to believe that AI is essential to humanity—or, at minimum, activism—reaching the next stage of its development.
I’ve long believed that activists in the future would use AI to automate social movement creation. Years ago, I wrote about the potential for activist AI in my book The End of Protest. In the section entitled Protest Bot, I speculated: "In the near future, the process of recruiting, training and deploying activists will be conducted by autonomous protest bots—computer programs augmented by artificial intelligence that spread the movement's memes and rituals." I then described in more detail how it could work. I recently revisited the topic in a short substack post here.
Of course, the existence of ChatGPT does not immediately solve the problem of how to do activism in 2023. As anyone who has conversed with the AI knows, ChatGPT will happily respond with plausible sounding but wildly false information. Ask it to brainstorm an activist campaign and it will do so, but that campaign probably won’t be very good.
And so, we are returned to the same place new movements always start from: a desire for change, a new tool and the time for experimentation. For Occupy Wall Street and Black Lives Matter that new tool was social media: Twitter and Facebook. For the next movement, that tool is AI.
So let’s move from critiquing AI to using AI—using it to create better and more resilient social mobilizations that orient hundreds of millions of people toward the highest imaginable goals. After all, if AI truly is a “pocket nuclear bomb” — well, that’s exactly how the activists in Ernest Callenbach’s Ecotopia got their utopia.