I spent last week in Pittsburgh for The SIGCSE Technical Symposium, that's SIGCSETS to you. Lots of talks, lots of sessions and lots of people.

We drove in on Tuesday since Batya had a workshop Wednesday afternoon. Devorah and I along with a cousin who was also at the conferences spent the day playing tourist. We hit the Warhol museum and also wandered by Randyland. Even made it to Carnegie Hall without practicing.

The opening keynote was given by Cecilia Aragon from the University of Oregon, the topic "What is Human Centered AI and why does it matter." The gist being that algorithms aren't free of bias regardless and that keeping the human element in CS is both critical and beneficial.

Here are some of what I felt were highlights. Some were core parts of the talk and others might have been a throwaway line but it still resonated with me or I otherwise thought it was memorable.

Early on, Aragon questioned if human centered was soft or easy with easy being erroneously perceived as either not rigorous or otherwise bad. She used to believe the easy is bad mantra due to CS's weed out culture but has since learned otherwise. This part of the talk was reminiscent of Felienne Herman's talk that's recently been making the rounds.

Aragon also talked about believing that ethics should be woven into all classes and not merely be a standalone even if required. This echoed with me as I've been saying this for years.

In another part of the speach, Aragon spoke on how she was working on a project to identify supernovas and how she moved it forward significantly in spite of two prior teams and team leaders failing to do so. She attributed the advancements to two factors. One was some technical, specifically algorithmic changes she implemented but the other was the human element. Aragon encouraged more collaboration and discussion and as this was an astronomy project, cross collaboration and discussion between the CS people and the astronomy people. It was that cross pollination that led to the technical improvements and was cited as an example of the "human side."

It was a good story and I agree but the sad part is that the example she gave shouldn't be an example of the "human side of AI" it's just how a good manager and good colleagues should work. Sadly it isn't always the case.

Another theme of the talk, and I mentioned this already is the fact that bias is already baked in. Our creations are biased because we are biased and that's something that we need constant reminders of and we have to do better.

A final point that was raised which I felt was particularly important given our current world situation, or more specifically, our current America situation was about a survey published by Stack Overflow. I forget the year and didn't copy down the specific numbers but you'll get the idea.

First question Aragon presented asked developers if they'd work on an ethically questionable. The majority said no which was a positive sign. Unfortunately, a later question asked something like what amount of responsibility a developer had when working on something unethical. In this case, the majority blamed management taking no responsibility for themselves. I was just following orders as it were. It was an old survey but I don't doubt the same mindset persists and it's greatly troubling.

Professor Aragon concluded the talk with recommendations on how we can do better.

Overall, a strong opening to the conference.