Now, we know chatbots aren’t a brand new thing, but we also know that what we have now, in the form of products like ChatGPT, and different. What is ChatGPT, then? You can go back to “Starting Out: What is Generative AI?” for a refresher of how chatbots work.

For many of us, generative AI hit our radars in a big way during November of 2022. For me, however, I remember showing students “text summarizers” back in the fall of 2019. We played with one of them, and we talked about the (not very good) results produced. I did not investigate the technology behind these sites (I can’t even find the original site I used as an example!); I simply focused on demonstrating how marginal and inconsistent the results were.

Fast forward to fall 2022 and the arrival of a free version of ChatGPT becomes ubiquitous. Now we had a tool that seemed to compose full essays out of whole cloth. “ChatGPT” has morphed into a “Kleenex” situation, as a catch-all term for generative AI, but that obfuscates the variety of tools and companies out there. And it is vitally important that we use precise language and evidence when we are peeling back these onion layers. There’s a lot of hand-waving and “Oh, it’s just a confusing black box situation” going on that keeps people unaware of the very real business industry behind these tools that seem to be everywhere.

People should be aware of the two major companies behind the two most popular chatbots: OpenAI (maker of ChatGPT) and Anthropic (maker of Claude). Of note, Anthropic was created by seven former OpenAI employees, including Dario Amodei (who was OpenAI’s Vice President of Research). I encourage you to close read the “About” pages of these companies. Many people have Microsoft CoPilot accounts as part of the 365 bundle; Google has Gemini.

When I was working on my dissertation (which is an exploration of serialized crime fiction published in the nineteenth century as a map of the ecological and Imperialist relationships established through the nineteenth-century British publishing industry), I realized that it was absolutely, fundamentally, critical uncover the origin stories of businesses, networks, and systems. The example I often use is Facebook — Mark Zuckerberg first experimented with negative user engagement with a website called FaceMash, where people could vote “hot or not” on the woman at Harvard (without their consent, of course). Who could have possibly imagined that the future of Facebook would include the amplifying hate prior to the Rohingya massacre, the Cambridge Analytica scandal, and conviction and penalty for harming children’s mental health? Talking about generative AI without investigating the industry that has created and deployed this technology leaves you with a massive gap in your understanding. For a long time, mass media coverage of OpenAI and Anthropic has been reporting what the developers are saying about the future of generative AI. That’s pretty much everywhere. It’s harder to find some of the more critical reporting about what is really happening, and so I have been collecting those articles and reports over the last four years:

Cracking open OpenAI:

Cracking open Anthropic:

Following the “AI” Industry