diff --git a/ai_preface.asciidoc b/ai_preface.asciidoc
index 18ba3839..a416f646 100644
--- a/ai_preface.asciidoc
+++ b/ai_preface.asciidoc
@@ -15,7 +15,7 @@ and who knows what it'll be able to do in a few years or even months' time.
 What we do know is that right now,
 AI is both insanely impressive and incredibly unreliable.
 
-Beyond being able to understand and respond to prompts in normal human language--
+Beyond being able to understand and respond to prompts in normal human language --
 it's easy to forget how absolutely extraordinary that is; literally science-fiction
 a few years ago--AI tools can generate working code, they can generate tests,
 they can help us to break down requirements, brainstorm solutions,
@@ -35,7 +35,7 @@ and undecipherable spaghetti code that spells a maintenance nightmare.
 === Mitigations for AI's Shortcomings Sure Look A Lot Like TDD
 
 If you read the advice, even from AI companies themselves,
-about the best way to work with AI, you'll find date that it
+about the best way to work with AI, you'll find that it
 performs best when working in small, well-defined contexts,
 with frequent checks for correctness.
 When taking on larger tasks, the advice is to break them down into smaller,