Learning about AI

Articles

• 'What is AI?' by Will Douglas Heaven, MIT Technology Review.
  This article dives into the different 'camps' around AI, pointing out how we lack a common understanding of what we are discussing when talking about this technology. Namely, some focus on the behaviour of this technology whereas others want to dive more into the mechanics of how this behaviour occurs. These different foci are how we can see the polarization regarding AI, there simultaneously being excitement and concern. As part of this discussion, the article considers the origins of artificial intelligence. While the article doesn't provide an answer to its titular question, I think it provides a nice broad overview to understanding the different sentiments around AI. 

• This opinion piece published through the Centre for International Governance Innovation about how discourse around AI has changed. I appreciate the last line summarizing thoughts around current stage of conversation, as I think it reflects my thoughts: "Sustaining this discourse will be key to getting the kinds of AI systems that we want: those that will advance, and not hinder, human flourishing."

• This Walrus article made me consider further possible advancements with AI. I appreciated how it highlighted the possible need for AI if wanting to maintain a certain level of productivity given declinining birth rates; while I'd heard of the fear around AI replacing workers, this need was a different angle I hadn't considered previously. Contemplating the Walrus article's idea of AI in robots (vs. apps) and how this may lead us to consider 'self' along with whether AI possesses this, paired with this Washington Post article on AI agents conducting tasks that highlights that '“With software, everything is written by humans' - I think this still points to the importance of recognizing what AI draws upon. What datasets are we training AI on, and with what biases or perspectives? 

• One of the most interesting discussions to me around AI is how it can 'replace' us, or in other ways, make us rethink what is 'human' (let alone human contribution).  The above two paired with this blog (which I admittedly just skimmed, but I take away from it that how humans contemplate or make decisions is important) seems like a fun place to think on it all. 

Resources

• Dr. Ethan Mollick is a management scholar who has been writing about teaching with AI, along with providing updates on AI's capabilities via his Substack. I'm not always the best at keeping up to date with Dr. Mollick's writing, but I appreciate what I've read and hope to check out his book. 

• Luiza Jarovsky, co-founder of the AI, Tech & Privacy Academy, writes a wonderful Substack column about AI governance. I skim this most of the time, but still enjoy getting glimpses of what is currently happenin. 

Developments

• Given concerns that people have with AI, specifically around how it is used and what biases it may reproduce, I'm glad when I read about researchers working on AI-related projects such as this hate speech detector that considers the contexts around certain words, such as community norms, to better understand a word's usage and connotations, or this tool that tells you what references that an AI-bot is drawing upon. 

  • Pairing the above with this LinkedIn post considering how AI could be used to identify microaggressions while helping to encourage an inclusive environment. I appreciate the note that we would need to consider AI's capacity to recognize when someone is being enthusiastic versus aggressive.  

• Something for me to look into more is 'retrieval-augmented generation AI', which is stated to ensure the sources that AI draws upon are verifiable. 

• I've only read a summary of options for how the AI economy could work in terms of crediting and compensating creators whose works are trained upon, but the summary referenced this paper. I'm partially posting this so I can find it later on to read. 

• A recent article discussed how ChatGPT advises women to ask for lower salaries than it would recommend to men for the same jobs. Note: I haven't read the actual paper. However, I would hope for accuracy with knowledge translation (ie., more publicly accessible) articles around the study. These results would be one of several examples of how AI can retain the biases we know humans possess.