May ChatGPT pray to God?
Will Theodore divine Samantha?
Is HAL ever going to stop blaming Dave?
Last Friday’s ChatGPT service, part of Deutscher Evangelischer Kirchentag
led 54 y.o. AI worker Heiderose Schmidt to reflect:
There was no heart and no soul. The avatars showed no emotions at all, had no body language and were talking so fast and monotonously that it was very hard for me to concentrate on what they said.
Thousands of AI workers have signed on to an open letter expressing ethical concerns about GPT-4.
My opening remarks towards our class homework question:
what kinds of AI does your place of business (POB) employ?
Answer: absolutely none.
POB HQ, August 2020: I rundown of our (then) 3 month LMS response to COVID. Primary takeaway: we weren’t even close to a beta (β) version.
Beginning of our last class, I closed our presentation with DS’s catherding slide:
My conclusion
when it comes to “data,”
we are all straying close to a fundamental (α) dimension,
ubiquitous-embodied-multi-modal
We are All at Level 1.
What last Friday’s German Evangelical service lacked was affect: ostensive gestures, embodied communication, self-efficacy seeking, connectedness and pleasure. As we implement AI into our lives, how can they help bring us a higher life?
RQ: What design factors for Cognitive Tutors and Teachable Agents can further Worker Education?
Here are some current Cognitive Tutors and Teachable Agents:
Of course, they can be used in more common digital management systems. For example, Co-Writer can write memos from highlighted spreadsheet cells. These education AIs are a subset of “machine learning.” Even middle school learners will master “machine learning,” if it will help them in a game. Visual tools can also facilitate machine learning.
Betty’s Brain is a good starting point to critique current designs of online learning platforms. 3 systems were designed for it:
- Learning by Being Taught (ITS) System
- Barebones Learning by Teaching (LBT) System
- Learning by Teaching with Self Regulated Learning (SRL) System
Betty’s Brain was able to guide confusion states by 8th graders towards delight and flow states. “At-risk” populations may be guided towards not disengaging. More importantly, when the SRL group had to expressly query Mr. Davis, an AI tutor, although they were initially slower to master content, moving forward their performance improved significantly. When all 3 groups were presented with new material, the SRL group performed best.
LearnLab was another early cognitive tutor applying the adaptive control of thought-rational (ACT-R) model. This model gathers dense data streams of student behavior and a large sample of students to greatly expand our knowledge of students’ mathematical cognition.
Another necessary design factor is successful integration into the school, into the classroom, and into a curriculum.