Impromptu: Amplifying Our Humanity Through AI Map my evolutionary progress Of course, centering users in this fashion also puts responsibil- ities on them. Fortunately, this is good both for the short- and the long-term. Especially the long. Today, LLMs like GPT-4 are clearly powerful but fallible, so there’s an obvious reason to stay attentive and hands-on. That has been a major theme of this travelog. As LLMs and other forms of AI evolve and grow more authori- tative and capable, however, it’s easy to imagine how we could acclimate to the convenience of machines that seem to literally do everything for us. After all, isn’t that the whole point of tech- nology? We go from painting on cave walls, to taking photos we develop in dark rooms, to Polaroid Instamatics, to Instagram’s automatically applied filters, to DALL-E 2. Or is technology’s ultimate goal to free oneself not from work, but for work? To help us do less—or do more? Until now it has always been the latter. I hope it continues to be the latter, not just for a few of us, but for most of us. I should emphasize that I use the word “work” here expansively. I mean human effort, human creativity, and human productiv- ity in all its variations—paid work, volunteer work, family work, artistic expression—whatever gives one purpose, meaning, and a sense of accomplishment and growth. I don’t want AI to erase that. I want AI to amplify that. But we have to be extremely intentional to achieve this outcome. There are versions of an AI future that could lead us down less beneficial roads—with the added challenge that these particular 218