Impromptu: Amplifying Our Humanity Through AI Large language models are similar to these technologies in that they can perform tasks that were previously done by humans. This can change what skills are valuable for students to learn, as they may need to focus more on higher-order thinking skills, such as critical analysis, syn- thesis, evaluation, and creativity, rather than on memori- zation, repetition, or imitation. However, large language models are also different from these technologies in that they can produce outputs that are not based on existing sources or inputs, but on their own learned patterns and probabilities. Students may need to verify, interpret, or modify the outputs of large language models, as well as understand the limitations, biases, or errors. Moreover, large language models can also learn from the feedback or interaction of students and teachers, which can create a dynamic and collaborative learning environment. Reid: In light of that, how would you recommend that American public education handle large language model tools such as you? GPT-4: I would recommend that American public edu- cation use tools such as me to augment and enhance the learning experience of students and teachers. I can offer personalized feedback, adaptive content, data anal- ysis, and interactive simulations that can help students develop their skills, as well as their curiosity and creativ- ity. However, I cannot replace the human elements of education, such as empathy, motivation, socialization. Therefore, I suggest that teachers use me as a comple- mentary resource, not a substitute, and that they main- tain a critical and reflective attitude towards my outputs and limitations. Reid: You say teachers should use large language mod- els as “a complement rather than a substitute,” and it 32