Ask The PhD Community

HigherEd AI Daily

June 28 – AI Credentials Multiply as Federal Oversight Grows

Sunday, June 28, 2026

Today's edition examines the rapid expansion of undergraduate AI programs across American campuses, federal action to slow a major model release, and candid career insights from a doctoral researcher entering the AI industry job market.

The Batch (DeepLearning.AI) — ACCESS

More Than 1,000 AI Programs Now Span U.S. Colleges and Universities

At least 1,000 AI programs now exist across approximately 584 U.S. colleges and universities, including 78 dedicated majors and 103 minors, according to April 2026 data from the Center for Inclusive Computing at Northeastern University. In 2021, only five institutions offered undergraduate AI majors. The rapid expansion reflects both surging student interest and institutional urgency to build workforce-ready graduates.

The programs range significantly in depth and philosophy. Carnegie Mellon University, which launched the nation's first Bachelor of Science in Artificial Intelligence in 2018, requires seven mathematics and statistics courses alongside dedicated modules in human cognition, perception, and ethics. At the other end of the spectrum, Drake University in Iowa offers an interdisciplinary Bachelor of Arts in AI designed specifically for humanities and business students, requiring only two mathematics courses and allowing flexible clusters across philosophy, English, psychology, and information systems.

Not all observers are enthusiastic. Some argue universities moved too slowly given AI's pace of development; others question whether specialized AI degrees come at the expense of broader computer science foundations. The University of Oklahoma Polytechnic Institute takes a practical applied AI approach, while Stanford University offers an AI concentration nested within its broader computer science framework.

Why it matters for campuses

The proliferation of AI programs raises pressing decisions for campus leaders: whether to build dedicated AI degrees, integrate AI competencies across existing majors, or pursue both approaches simultaneously. Faculty senate processes and curriculum approval timelines, typically measured in semesters, are poorly matched to a field where foundational frameworks shift within months. Institutions that act strategically now, including investments in interdisciplinary models like Drake's that lower math prerequisites, may be better positioned to serve diverse student populations rather than only those with strong quantitative backgrounds.

Read More

TLDR AI — GOVERNANCE

White House Issues Administrative Request Asking OpenAI to Delay Its Next-Generation Model Release

The White House has issued an official administrative request asking OpenAI to delay public deployment of its next-generation frontier model. The government's stated concerns center on national security and structural safety considerations, with officials pushing for an extended red-teaming window to audit the system's advanced cyber-capability execution limits and automated social manipulation vulnerabilities before it reaches consumers and institutional users.

The request represents one of the most direct uses of executive authority to intervene in the deployment timeline of a commercial AI system. It signals that federal agencies are actively monitoring not just AI policy in the abstract but the specific readiness of individual frontier models. The action arrives as multiple institutions, including universities, are integrating frontier AI tools into research and instructional workflows without formal federal guidance on acceptable use.

The concerns flagged specifically around automated social manipulation vulnerabilities are worth noting in a campus context; universities are environments where large volumes of students interact with AI-assisted advising, writing support, and research tools at scale, creating meaningful exposure if similar risks are inadequately evaluated in commercial deployments.

Why it matters for campuses

Academic leaders and campus AI governance bodies should take note of how the federal government is framing frontier model risk. Institutions that have deployed or are piloting frontier AI tools for research, advising, or instruction are operating against a regulatory backdrop that is becoming more active. Campus AI policies should anticipate potential disruptions to model access and build contingency approaches. This is also a timely moment for faculty governance to weigh in on how institutions vet and adopt AI systems whose safety profiles remain under active federal review.

Read More

TLDR AI — RESEARCH

A Fifth-Year PhD Student's Industry Job Search Reveals Surprising Realities About AI Career Readiness

A fifth-year doctoral student at Brown University recently published a detailed account of a research scientist job search in Silicon Valley, and the findings challenge common assumptions about how advanced degrees translate into AI industry careers. The most surprising finding: only one or two research papers truly mattered in the hiring process, regardless of total publication count. Depth on a small number of projects carried more weight than breadth across many.

The interview process proved far more varied than expected. Different organizations ran substantially different rounds, covering areas well outside the candidate's core expertise, and timing emerged as a critical factor. Many opportunities arose from alignment between a company's immediate needs and the candidate's background rather than from an initial strong match on paper. The author notes that many of the most productive interviews tested how well-rounded the candidate was as an AI researcher, not how deeply specialized in a single domain.

The account is candid about the role of luck, network, and external circumstances; and it resists the tidy narrative that a strong publication record in a narrow specialty reliably produces industry research roles. The ideal graduate heading into research scientist positions appears to be someone who can move across problems, communicate reasoning clearly, and adapt to interview formats that range widely.

Why it matters for campuses

Faculty who advise doctoral students considering industry careers should incorporate these realities into mentorship conversations early, not just in the final year. Graduate programs that structure curriculum and advising primarily around deep specialization may be underserving students who will enter a job market that rewards adaptability, communication, and broad research fluency. Departments might consider whether capstone experiences, external collaborations, or interdisciplinary coursework better prepare graduates for the research scientist roles that are increasingly the destination for many AI-adjacent PhD graduates.

Read More

Tool of the Day

GLM-5.2 (Z.ai / THUDM)

GLM-5.2 is a newly released open-weights AI model available under an MIT license, offering 753 billion total parameters with a 1-million-token context window and benchmark performance that rivals top commercial models including GPT-5.5 and Claude Opus 4.8. The model weights are freely downloadable via HuggingFace, making it accessible to researchers and institutions regardless of licensing budget. API access is also available at significantly lower cost than comparable closed models.

For educators and researchers, the 1-million-token context window is the standout feature: it allows an entire textbook, a semester of student essays, or a full qualitative dataset to be processed in a single query, opening practical possibilities for large-scale feedback, synthesis, and research analysis that were previously impractical with smaller context models.

Try it: Download GLM-5.2 via HuggingFace and load your complete course readings for the current semester, then prompt the model to identify recurring conceptual tensions across the texts and generate three discussion questions that would help students synthesize the material before a final exam.

Visit GLM-5.2

Have a great learning day!

Dr. Ali Green

Sources for This Edition

The Batch (deeplearning.ai)
TLDR AI (tldr.tech/ai)
Center for Inclusive Computing, Northeastern University (northeastern.edu)
HuggingFace / THUDM (huggingface.co/THUDM)

askthephd.com
 | 
askthephd.substack.com
 | 
Unsubscribe

HigherEd AI Daily; Curated by Dr. Ali Green