The 4D Framework for pK-12 educators — Transcript

Full transcript

1. 0:01[snorts]
2. 0:02[music]
3. 0:07>> We want you to build lasting AI fluency.
4. 0:10That means giving you the tools to use
5. 0:12AI efficiently, ethically, and safely,
6. 0:15no matter what challenges you might
7. 0:16face. In this video, we'll walk through
8. 0:18the 4D framework, four interconnected
9. 0:21competencies that, when combined,
10. 0:23transform how you work with AI. To help
11. 0:26you apply this framework, we'll look at
12. 0:28two modes of interaction with AI. The
13. 0:30first mode is likely the one you're
14. 0:32familiar with, how to engage with AI
15. 0:35effectively on a day-to-day basis. This
16. 0:38is the inner loop of description and
17. 0:40discernment. You describe what you want
18. 0:42AI to help you with, and then you
19. 0:43discern if it meets your expectations.
20. 0:46This is a critically important skill set
21. 0:48when learning AI, but it's not
22. 0:50sufficient on its own. In addition to
23. 0:52everything that happens within your
24. 0:53interactions with AI, there's also an
25. 0:55outer loop of delegation and diligence.
26. 0:58This is everything that happens in your
27. 1:00workplace, community, and life around
28. 1:02the use of AI. Should you be using AI at
29. 1:04all? And if you do, what are your
30. 1:07ethical responsibilities to ensure
31. 1:08you're proceeding intentionally and
32. 1:10responsibly? We'll be doing a brief
33. 1:12overview of all of this today, but if
34. 1:14you'd like a deeper understanding, check
35. 1:16out the AI Fluency Framework and
36. 1:18Foundations course on Anthropic Academy.
37. 1:21Let's begin with a close look at the
38. 1:23inner loop of description and
39. 1:25discernment. Again, this happens during
40. 1:27your day-to-day interactions with AI.
41. 1:31Description is how we refer to
42. 1:32communicating effectively with AI
43. 1:34systems. Good prompt matters, but so
44. 1:37does everything around it. The context
45. 1:39you provide, the examples you share, how
46. 1:41you refine and redirect the
47. 1:43conversation.
48. 1:44It all starts with defining what you
49. 1:46want, your outputs, format, audience,
50. 1:48and style.
51. 1:50Instead of write a summary of this
52. 1:51report, try getting specific.
53. 1:54Summarize this report for an
54. 1:56eighth-grade audience, focusing on the
55. 1:57three highest impact findings in under
56. 1:59300 words.
57. 2:01But you can also define how the AI
58. 2:03approaches your request. Think of it
59. 2:05like giving instructions to a
60. 2:06collaborator specifying the steps, the
61. 2:09order of operations, or the reasoning
62. 2:11approach you want the AI to follow.
63. 2:13And you can shape the AI's behavior
64. 2:15during your collaboration as well. Do
65. 2:17you need a critical reviewer who pushes
66. 2:20back on weak arguments, a brainstorming
67. 2:22partner who builds on every idea, a
68. 2:24fact-checker who flags uncertainty? You
69. 2:27can ask for all of it.
70. 2:29In our research, we found that
71. 2:30description skills are the ones people
72. 2:32pick up most naturally. Things like
73. 2:35specifying what you want, giving
74. 2:36examples, and refining your requests.
75. 2:39Most people are already doing some
76. 2:40version of these, and that's a great
77. 2:42starting point. As you iterate and
78. 2:44collaborate with AI, it's likely you'll
79. 2:46get more comfortable and effective at
80. 2:48description.
81. 2:50Discernment, the companion to
82. 2:51description, asks you to thoughtfully
83. 2:54and critically evaluate your AI
84. 2:56collaborator. Start with the output
85. 2:58itself. Are these statistics accurate?
86. 3:01Does this language reflect how your
87. 3:02audience actually talks about this
88. 3:04topic? Then look at how AI got there.
89. 3:07Did it consider all relevant factors?
90. 3:10Importantly, discernment doesn't just
91. 3:12mean accepting or rejecting AI outputs.
92. 3:14It involves iteration to get where you
93. 3:17want to go.
94. 3:18And while most people describe what they
95. 3:20want fairly naturally, discernment is
96. 3:22much rarer. In our research, skills like
97. 3:25questioning AI's reasoning, checking
98. 3:27facts, and identifying missing context
99. 3:29were among the least common behaviors we
100. 3:31observed. And that's where there's room
101. 3:33to grow.
102. 3:34It can be an easy step to miss, but it's
103. 3:36critically important. It's what drives
104. 3:38true AI collaboration. That's why
105. 3:41description and discernment work
106. 3:43together in a loop. You describe what
107. 3:46you need, evaluate what you get, and
108. 3:48then refine your description based on
109. 3:50that evaluation. It's like working with
110. 3:52a human teammate. You build a shared
111. 3:54understanding through conversation.
112. 3:57Using AI well in day-to-day interactions
113. 3:59is only one piece of the puzzle. It's
114. 4:01critical to understand all the ethical
115. 4:03and social implications of AI use in
116. 4:05your context. This is where the
117. 4:07delegation diligence loop comes in.
118. 4:09Delegation is deciding what work should
119. 4:11be done by humans, what work should be
120. 4:12done by AI, and how to distribute the
121. 4:14tasks between them.
122. 4:16Good delegation starts with
123. 4:18understanding the work itself. Before
124. 4:20drafting an email, ask yourself, is this
125. 4:22a routine status update where AI can
126. 4:24handle a first draft, or is this a
127. 4:26sensitive message to a family about
128. 4:28their child's progress where tone and
129. 4:30nuance need to come directly from you?
130. 4:32You also need to understand your tools.
131. 4:35If you're working with confidential
132. 4:36data, for example, you'd want to
133. 4:38consider the privacy and security
134. 4:39features of the AI system you're using.
135. 4:42When you understand both the work and
136. 4:44the tools, you can thoughtfully split
137. 4:45tasks to leverage the strengths of each.
138. 4:48We pair delegation with diligence,
139. 4:51taking responsibility for how you use
140. 4:54AI.
141. 4:55That means being thoughtful about which
142. 4:57AI systems you use and how you interact
143. 4:59with them. Maybe you use AI to draft a
144. 5:01project proposal, but you're intentional
145. 5:03about which sections it works on and
146. 5:05which require your own expertise.
147. 5:08It means being honest about AI's role in
148. 5:10your work with the people who need to
149. 5:12know.
150. 5:13Your colleagues and stakeholders deserve
151. 5:15to understand when and how AI has been
152. 5:18involved. And it means taking
153. 5:20responsibility for verifying and
154. 5:22vouching for the outputs you use or
155. 5:23share. You check every claim, confirm
156. 5:26accuracy, and make sure any final
157. 5:28deliverable generally represents your
158. 5:30goals and standards. You're always
159. 5:32accountable for the final result.
160. 5:34Delegation and diligence work together
161. 5:37as a loop. The thoughtful choices you
162. 5:39make about what to delegate to AI must
163. 5:42be matched by ongoing responsibility for
164. 5:44how how use it. And your diligent
165. 5:47practices inform smarter delegation
166. 5:49decisions over time.
167. 5:50The four D's come alive when you use
168. 5:53them together, creating a new syllabus,
169. 5:55use delegation to decide what AI handles
170. 5:58versus what you bring yourself.
171. 6:00Use strong description to guide the AI's
172. 6:02work, apply discernment to evaluate the
173. 6:05results, and practice diligence
174. 6:07throughout by choosing appropriate
175. 6:09tools, being transparent about AI's
176. 6:11role, and taking responsibility for
177. 6:13accuracy. This framework is about making
178. 6:15you more effective at the work that
179. 6:17matters. Work that requires human
180. 6:19judgment, creativity, and deep
181. 6:21understanding of your context.
182. 6:23Throughout this course, you'll apply
183. 6:24these competencies to real challenges,
184. 6:27creating differentiated instructional
185. 6:29materials, analyzing data while
186. 6:31protecting student privacy, and building
187. 6:33lesson plans. The skills you take away
188. 6:35will serve you as AI tools continue to
189. 6:37evolve. You can learn more about AI
190. 6:40fluency in Anthropic Academy, and we'll
191. 6:42continue to share our research on this
192. 6:43topic on Anthropic's blog.
193. 6:47>> [music]