Superagency

Superagency offers a cautious and optimistic look into our AI future. Hoffman and Beato offer four main classifications to sort stances towards technology and specifically, AI. Doomers think only worst-case scenarios, many jobs will be lost. Gloomers are highly critical of both AI and Doomers; they are concerned with potential job losses and also in the loss of individual agency. By contrast, Zoomers think all the gains of AI far exceed the negatives; new and different jobs will be created. Finally, Bloomers, where the authors stand, are optimistic about AI’s promise. They welcome progress while acknowledging its complex challenges.

Superagency does not shy from pointing out challenges surrounding AI, however it does so through a Bloomer viewpoint.  For example, how might the adoption of AI for mental health clinicians go right? Personally, I found this section a refreshing and interesting read amid reports of the massive failures of chatbots for mental health and amid concerns of LLMs providing misinformation. The authors identify different stages that clinicians can explore the use of AI. To begin, they can use AI to help with record creation and transcript generation; two tasks that are tedious and necessary. In the second stage, clinicians can use AI to assist in reviewing transcripts and provide some basic analysis to spot trends and adherence to therapies to flag for more personal detailed review. Finally, a clinical LLM can be trained on EBPs. Overall, it can help train more practitioners, and function as a support for them by aiding in managing a (potentially larger) case load more efficiently. Finally, it will allow many who are currently underserved  access to necessary and affordable care.

Record scratch—how can this be good? Humans make mistakes every day. So how confident can we be in an LLM? How does AI earn that trust? One approach is regulation, though its exact shape remains undefined. Hoffman explores benchmarks and tests to evaluate model performance as well as tools such as the Chatbot Arena.

Another area in which AI has a rosy future is because we have to constantly learn new skills to keep up with new technologies to stay innovative. LLMs are useful for being infinitely patient with beginner mindset. They can help a learner get on board quickly, and unlike in person teachers, don’t mind if a query is set aside for hours (or days) to be returned to at another time.

Every new technology carries risks. Critics often call novel tools¹ dehumanizing and limiting. But they often end up democratizing and liberating. Jobs aren’t simply taken away, they transform into more creative, strategic roles.

If you’re eager to dive into a Bloomer’s hopeful vision for AI’s potential, Superagency offers clear and engaging insights. If you’re searching for bright examples of AI’s life-changing promise, this book will inspire and inform.

1. For example of a novel tool, see https://www.youtube.com/watch?v=pQHX-SjgQvQ ↩︎