Tuesday started with two more sessions. First up was "Finding Balance with AI" led by fellow New Yorker Matt Carlberg.

The session was about how Matt has worked to find the same balance many teachers are struggling with - teach the students to use generative AI effectively and safely while also keeping a handle on issues like cheating. The session was a mix of Matt sharing his practices and small group discussions.

Matt described assignments as being AI closed - no AI allowed, AI Open - where students might use AI in an unlimited fashion, and hybrid between the two extremes. So, a programming assignment done in class under direct teacher supervision might be closed whereas homework would be open since there's not real way of controlling whether or not a student is using AI at home.

One particularly interesting point was when Matt described how he runs class lab assignments. He uses Pickcode. Pickcode is an online educational IDE built by a small team of developers who work closely with teachers. In fact, a teacher request led to a Pickcode feature critical for Matt's practic. Pickcode allows a teacher to lock and unlock assignments based on time. That means that a student can only access a class lab assignment during class time - critical if you want a level playing field for all students.

Matt also described his final class project.

The first part was AI closed - all done in class. Students designed and built an RPG battle game. Part 1 was all done in class and combined written and coding parts. It was worth 60 points.

Part 2, the intermediate level was AI open but students had to record their AI sessions. This was worth 20 points.

For the final 20 points, students could optionally complete the advanced part of the assignment and that was to use AI to learn something new - in the case of this assignment the thing to learn was inheritance. Ten of the points were for completing the task and the final ten if the student took an optional in class on paper quiz.

It was an interesting assignment and an interesting shot at both using and controlling the use of AI.

None of us really know what the best approach will be but with people like Matt experimenting, refining and sharing their practices, we'll get there.

My second session was "Beyond the foundation - planning with the new HS specialty standards."

The big announcements this year at CSTA were all around standards.

In addition to releasing their new PK - 12 standards they also released teacher standards. I'll write about those at some later date.

For he PK - 12 standards, in addition to what they are calling foundational standards - the stuff all kids should know, they also released specialty standards.

Seven more areas for say, electives after the "foundational standards" are covered.

What are they?

1. Artificial Intelligence
2. Cybersecurity
3. Data Science
4. Game Development
5. Physical Computing
6. Software Development
7. X + CS (another subject plus CS)

We didn't look into these in depth but we did look over the web site (https://csteachers.org/pk12standards/) which has a really slick interface and great filtering capabilities.

While I'm not big on "the standards" (more on this in the next post), these are a nice addition. They don't, in my opinion, provide standards as in "all students should know all of this of they study, say AI" but rather these look like nice references for teachers who want to build courses or experiences in these areas.

The existence of these standards can also provide "cover" for a lone teacher trying to justify the creation of a course.

Then we had our second keynote, Cynhis Breazeal talking about "Holistic AI Literacy" - how we might think about AI. It was fine and the speaker was engaging, but it didn't really do it for me.

The day closed with two receptions. One off site, sponsored by Juicemind. I didn't attend that one. The other back at the venue split up by grade band - elementary, middle, and high school. I stopped by the HS one. It was pretty cool with the organizers provided "bingo boards" to encourage people to meet one another.

So, much like day 1 - a couple of good sessions, an uneven keynote and lots of good conversations.

Day 3 and a few more thoughts still to come.