WAC LIAISONS - AI Recent News

Leslie found the following texts helpful for maintaining a broad AI cultural currency.  Her choices address the economy, racism, medicine, new AI tools, and more.  The resources are orderd chronologically to ease updating, which will happen at least quarterly.  Enjoy!

Resources

Keen, M.  AI Trends for 2025.  IBM Technology.  Accessed 9 Jan. 2025.

• Keen, an IBM Fellow, predicts the top seven AI trends for 2025.  #7, "Human-in-the-loop augmentation" seems most promising for educators.

Elliott, C. (18 Dec. 2024).  AI Is Changing How We Study Bird Migration.  MIT Technology Review.  Accessed 9 Jan. 2025.

• AI tools are helping ornithologists study migrating birds, recording and analyzing more data, even at night, with AI.

Alexander, B. (23 Nov. 2024).  More Opposition to AI.  AI, Academia, and the Future.  Substack.  Accessed 25 Nov. 2024. 

• Futurist Bryan Alexander provides a timely overview of categories of opposition to AI, stemming from culture, including "AI and Human Relationships," "AI against Artists," and "AI and Kids."

Roth, E. (14 Nov. 2024).  NASA's AI Earth Copilot Will Take Your Questions about Our Planet.  The Verge.  Accessed 18 Nov. 2024.

• In collaboration with Microsoft, NASA is building an "AI Earth Copilot" which will make "it easier to access and understand" NASA's massive data collections.  Includes a video demonstration.

Pomfret, J. and J. Pang. (1 Nov. 2024).  Chinese Researchers Develop AM Model for Military Use on Back of Meta's Llama.  Reuters.  Accessed 18 Nov. 2024.

• Meta releases the "weights" of its Llama AI model, allowing anyone to build on it for their needs.  China's PLA used it to create a military tool.  Meta says it was "unauthorized" to do so.  This may inform future efforts to regulate frontier AI models.

Hadhazy, A. (28 Oct. 2024).  Can AI Improve Medical Diagnostic Accuracy?  Human-Centered Artificial Intelligence.  Stanford University.  Accessed 18 Nov. 2024.

• Spoiler alert:  yes.  In Stanford's study of 50 physicians and ChatGPT-4, the AI diagnosed patients' health issues with a median score of 92, while physicians scored between 74 and 76 (non-AI users and AI users, respectively).  The team speculates that tools dedicated to supporting physicians' work are needed, as is training for physicians in how to best "collaborate" with the tools.  Original study in JAMA.
  

Roose, K. (23 Oct. 2024).  Can A.I. Be Blamed for a Teen's Suicide?  New York Times. Accessed 18 Nov. 2024.

• Kevin Roose interviews the mother of a boy who became obsessed with an AI companion and eventually committed suicide.  What obligations do tech companies have to install and test guardrails in their AI products?

Roose, K. and C. Newton (Hosts).  (27 Sept. 2024).  Meta's Race for Your Face, Google's Hit AI Notebook, and HatGPT (Audio Podcast).  Hard Fork.  New York Times. Accessed 2 Oct. 2024.

• This whole episode is great.  That said, the section in which Roose and Newton discuss the possibilities for Google's "NotebookLM" tool with author Stephen Johnson will inspire you to create your own notebooks--and teach your students to use them to study for your class--ASAP.

Roose, K. and C. Newton (Hosts).  (20 Sept. 2024). "OpenAI's Reasoning Machine, Instagram Teen Changes and Amazon R.T.O Drama."  (Audio Podcast).  Hard Fork.  New York Times. Accessed 23 Sept. 2024.

• Listen to learn about OpenAI's expensive new tool, OpenAI 01, a "reasoning" tool that answers queries using a process called "Chain of Thought."

Miller, K.  (3 Sept. 2024).  "Covert Racism in AI:  How Language Models Are Reinforcing Outdated Stereotypes."  Human-Centered Artificial Intelligence.  Stanford University.  Accessed 23 Sept. 2024.

• Miller reports on an article in the journal Nature that found that efforts to reduce overt bias in genAI outputs have actually increased the tools' covert biases.
  

Khullar, D.  (2 Sept. 2024).  "How Machines Learned How to Discover Drugs."  New Yorker.  Accessed 23 Sept. 2024.

• Khullar, a medical doctor, describes how researchers are teaching AI to identify promising compounds that may help humans combat antibiotic resistance in deadly microbes.
  

Gerken, T.  (28 Aug. 2024).  "AI Lets Us Cut Thousands of Jobs--but Pay More."  BBC.com.  Accessed 23 Sept. 2024.

• Financing company Klarna is laying off thousands of employees, admitting it's possible because they're "using AI in marketing and customer service."  Klarna leadership says the government needs to support these and other thousands who lose their jobs to AI.
  

Bhatia, A. (25 Aug. 2024).  When AI's Output Is a Threat to AI Itself.  The New York Times. Accessed 9 Jan. 2025.

• When AI is trained on its own creations, its output converges toward the most average results, leading to inaccurate, useless outputs.  Watch the interactive portions and take hope that current AI models, which use probabilities to produce outputs, will not exceed human abilities.