SAIL: Empowering Learners
AI is a field that never sleeps. A constant hourly deluge of new trends and new advances. The State of AI Report 2022 provides an overview of the pace of change. It's a solid overview, in line with Mary Meeker's state of the internet reports. I was interested to see how vested universities remain in AI spinouts (slide 67, but balanced with slide 82 showing industry is the holder of large data sets, though decentralized research collaboratives are growing slide 84) and general economics (slides 70-78).
Conferences and Webinars
GRAILE is now running its 3rd Annual Conference on Empowering Learners for the Age of AI. Outstanding keynotes have been posted as well as a diverse range of panels.
GRAILE will be holding pre-conference meetings pre-LAK13 in Arlington/Dallas March 13, 2023. If you are in the area, please let us know!
Did you miss our webinar on learner profiles? If so, the recording is now available here.
For those in the USA targeting NSF or IES grants, you've likely heard about SEERnet. It's worth spending some time looking through the site. Data, data aggregation (lakes), and various platforms that feed that lake, provide the next level of research opportunities. ASU's
AI & Learning/Education
This is an argument that we will be having many times going forward, especially as AI advances in its performance on many human cognitive tasks. What should be taught when AI performs close to human levels. Or, increasingly, exceeds them.
AI Technology and Arguments
Gary Marcus continues to throw cold water on the hype that comes out of "pure AI labs" such as DeepMind. There is a view that the path forward with AI to eventually get to artificial general intelligence (AGI) called "scaling maximalism". The idea here is that we are happily on the right path with the right architecture and we just need to scale. Marcus disagrees.
Can AI be sentient? This is starting to feel like pub conversations for philosophers. The intent isn't to answer the question. The entire value is in the asking.
Ethics and Rights
The White House released a Blueprint for AI Bill of Rights. Something like this should be on the minds of all AI developers in education. There are important human rights at the center of every algorithm.
Data forms the foundation of models. Many of the language and image data sets that underpin media-prominent AI tools rely on data that has been created and share by end users who were not paid for their data or aware that it might be used for future models. A group of GitHub users are suing Microsoft because their code was used to train AI pair programming tool Copilot.