Conclusion: At the Crossroads of the 21st Century Over the course of this travelog, I’ve highlighted the intense engagement that LLMs like ChatGPT and GPT-4 induce with their users. Right now we see that playing out in interest- ing ways. People who think it’s incredibly reckless to put AI’s power into the hands of millions are highly engaged in expos- ing the shortcomings, biases, and malfunctions of these new tools. People frustrated that AI developers are taking steps to reduce harmful or toxic outputs are highly engaged in finding ways to overcome those constraints. The work of both these groups is extremely valuable to a third group, of which I’m a member. This is the group that aspires to design and use AI in ways that, over the long term, broadly empower all human beings, not just a favored few, to amplify their abilities, opportunities, and human agency. That’s a huge ambition. To achieve it, everyone must contribute to the effort—including the skeptics, including the ones trying to undo a product’s safety engineering. So I hope their high engagement continues. The road to utopia is paved with failure and loss To achieve this ambition, we must also accept both the inevita- bility of loss and failure and the necessity of regulation. Harnessing the power of fire brought cooking and the hearth, but it also led to arson and regulations about where you can place a BBQ grill, especially if you live in an apartment. The wheel revolutionized transportation, agriculture, and engineer- ing, but it also paved the way for high-speed car crashes and traffic lights. In the U.S. alone, more than 400 people over the age of fifteen injure themselves in bathtubs or showers each day. So we have 211