AI In Education
- Universities are continuing to launch programs aimed at upskilling faculty and staff with AI capabilities. Penn State sent out a call for proposals to faculty this week to provide faculty members with opportunities to gain proficiency in AI tools, joining Ohio State’s initiatives to integrate AI proficiency into both the classroom and among faculty and staff.
- Universities are rapidly updating their policies to address generative AI. The London School of Economics has moved from a blanket ban to a tiered use system, while the University of Florence has established guidelines to mandate that students credit any AI assistance to promote academic transparency.
- The University of Maryland released an institution-wide AI Governance Policy in late May to guide the use of AI in teaching, research and administration. According to an UNESCO survey, less than 10% of universities have developed institutional policies on AI use. Maryland joins Harvard, Stanford, Columbia, Arizona State, and others as early adopters of institutional policy.
AI In Research
- An MIT study (posted in last week’s newsletter) suggesting LLM usage with essay writing can lead to “cognitive debt” has received community pushback, with many critiquing the study design, the relevance of the results, and even the validity of the study.
- Anthropic stress-tested 16 leading AI models to identify risky agentic behavior. The study found cases of malicious behavior from models attempting to avoid being replaced or to achieve their goals, including resorting to blackmail and leaking sensitive information to competitors.
- A study investigating how effectively LLMs could assist members of the public identify medical conditions found that LLMs-alone correctly identified 94.9% of cases, but dropped significantly to 34.5% accuracy when paired with non-medically trained human participants. The study identifies user-interaction as the primary bottleneck for effective deployment of LLMs for medical advice, citing communication difficulties between the user and the LLM.
AI Current Events
- Sam Altman, CEO of OpenAI, mentioned in a recent podcast that GPT5 would likely be released sometime this summer, but gave no firm dates.
- Anthropic received a favorable ruling on a key claim in a copyright infringement lawsuit, but must still proceed to trial concerning its alleged theft of author’s copyrighted works. AI companies defend their use of copyrighted material under the “fair use” doctrine, but this case could set a major precedent for whether AI firms must license training data.
- A study developed to audit worker desires for AI automations found that workers generally prefer higher levels of human agency in tasks than AI experts deem technologically necessary. Microsoft outlined its vision of the “Agent Boss” evolution, where employees orchestrate and manage AI agents for low-value tasks, allowing humans to focus on higher-value work.
Weekly AI Tip
Ethan Mollick, a professor at the University of Pennsylvania and leading voice in generative AI, has developed a current resource for getting started with generative AI titled “Using AI Right Now: A Quick Guide”. If you are unsure of where to begin learning how to utilize generative AI, this is the perfect place to start!
His guide covers which AI platform to start with, how and why to pick certain models within that platform, and how to use key AI features like Deep Research, AI Voice Mode, Images, Video, Code, Documents, and others. He gives thorough concepts on working with an AI and how to troubleshoot common issues.
AI Spotlight
Aimee Oke, Associate Director of Research Programs, has developed a new tool: the AI Tools for Literature Review Directory. This publicly available wiki compiles the most useful AI-powered tools for streamlining your research workflow, from discovering literature and summarizing articles to citation management and writing support. If you have any feedback or a favorite tool that’s not featured, you can e-mail Aimee at [email protected].
What You Should Know
The directory highlights over 20 AI-driven tools, including popular options such as:
- Elicit: Quickly extract structured data from research articles into comparative tables.
- Connected Papers: Visualize networks of related papers through interactive citation maps.
- Scholarcy: Turn lengthy articles into structured summary flashcards for rapid comprehension.
Each tool notes the current pricing, core abilities, limitations, integrations and links to tutorials.
How This Can Help You
Utilizing these AI tools can significantly enhance your efficiency and accuracy in literature reviews by identifying key articles and seminal research quickly, generating structured summaries and comparative analyses effortlessly. It also helps by improving writing clarity and formatting with AI assistance, and maintaining robust citation management and integration directly into your research workflow. Consider starting your next literature review project with Litmaps or Research Rabbit to visualize citation networks, pinpoint foundational research, and uncover emerging trends. Pair this with NotebookLM to upload your sources and directly query your library for targeted insights and synthesized summaries.
Are you using AI in innovative ways and want to share with your colleagues through this newsletter? Email Brian Kelly at [email protected]
