“Write me an article about AI and job insecurity.” Each day, OpenAI’s ChatGPT receives 2.5 billion prompts, including ones like this. As of Q2 2025, Claude, an AI designed by Anthropic, is now embedded in 60% of Fortune 500 companies’ productivity suites. With AI’s growing presence in our world, concerns surrounding its use are also increasing, with a growing number of white-collar workers predicting job insecurity. This trend has seemingly followed at some of the top institutions, with 54% of Harvard College students in the Harvard Political Review’s Fall 2025 Campus Poll citing AI as a threat.
This perception is not new. An April 2023 study found that 41% of workers between the ages of 18 and 25 expressed concern about AI making their jobs obsolete. Yet, this fear does not stop AI from growing. Since ChatGPT’s launch in November 2022, the weekly user count has grown to 800 million people. This trend is particularly alarming among students at top universities, such as Harvard, where high AI usage has led to the creation of university-wide use guidelines, and 56% of students in the poll have cited “often” or “very often” for their frequency of use.
These conflicting perspectives create a mixed legacy for AI: what effect will AI have on careers for the newest generation of college students, and how will this impact the way college students are educated?
Each university approaches this conundrum differently. While some mandate strict AI use standards, others encourage students to become proficient to optimize their future career opportunities. Both approaches inherently have their own benefits and drawbacks, forcing top universities to consider the boundaries of a legitimate education.
At Yale University, students agree with renowned computer scientist Kai-Fu Lee’s idea that “AI will not take jobs, but humans who learn AI will.” To ensure their students are well-equipped, Yale will invest $150 million between 2025 and 2030 in AI research, security, appropriate use, and education. While Yale has their own guidelines for AI usage, focused on protecting confidential information, professors are given leniency to set rules regarding permissible use in their own classrooms. Still, Yale’s paramount focus is teaching students to become responsible users through courses and outside resources, which may ultimately push students to depend on AI for critical thinking.
Dartmouth University is pursuing a similar track. In the spring of 2025, Dartmouth piloted new “AI literacy” content in writing seminars. In October 2025, Dartmouth formed the AI Faculty Leadership Committee to embrace and harness AI, and in November, researchers designed an AI chatbot for students in the Geisel School of Medicine to increase tailored learning with fewer hallucinations. Much like Yale, Dartmouth University still leaves AI usage up to professors. Despite being the place where the term “artificial intelligence” was first coined, Dartmouth remains the most behind among top institutions, posing the question of whether it is ready to deal with AI’s fast growth.
Similarly, Columbia University encourages professors to set their own guidelines in the syllabus regarding permitted uses of AI, while simultaneously investing in various AI-based tools and research surrounding the impacts of AI across various sectors through their initiative Columbia AI. On top of their research, Columbia has also created a new AI minor, currently limited to engineering majors, pushing students to learn how to effectively use AI while also teaching about the ethical drawbacks. Columbia, in comparison to Dartmouth, has an extremely large focus on the ethics of AI and its usage.
While Dartmouth and Columbia both outright ban AI usage if it is not mentioned in the syllabus, Stanford University treats AI usage “analogously to assistance from another person.” Historically, Stanford has pushed for strong education, research, and practice in the field, with the founding of the Stanford AI Lab in 1963. Even today, Stanford continues to innovate, piloting a school-wide platform called the Stanford AI Playground to allow users to securely try AI. On top of their other programs, to explore all angles of AI, Stanford Human-Centered Artificial Intelligence focuses on policy interventions to “improve the human condition.” Still, Stanford’s course policies of treating AI as a peer may be encouraging unethical use of AI to circumvent the work required to earn a degree.
Like Stanford’s AI Playground, Harvard offers AI Sandbox, a confidential program that allows users to explore generative AI without their data being used to train large language models (LLM). Harvard is also exploring AI from a variety of different angles, with researchers looking into its impact on health, education, art, economics, policy, and more. Like Dartmouth and Columbia, Harvard leaves allowed generative AI usage up to professors; however, instead of focusing on students, Harvard is leading trainings, office hours, and workshops to teach staff how to integrate AI into their classrooms and their lives.
While these top institutions have different focuses and guidelines for AI usage, they can all agree that the future job market will be flooded with AI. To ensure their students remain competitive in the job market, each school is pushing for AI integration in course curricula, secure AI platforms for experiential learning, and funding for research. These changes are allowing the top institutions to fight fears of job insecurity one step at a time.
Still, can students who use AI at these institutions truly earn their degree? Is treating AI like a peer or allowing teachers who may not know the bounds of AI to create guidelines truly ethical learning? For 45% of students in our voluntary poll, AI means non-coding assignment help, for 49% it means summarized readings, and for 26% it means essay drafting. Gradually, we are seeing a student shift away from critical thinking and toward a dependency on AI.
Soon, top institutions will be forced to face the reality of our generation: increasing AI usage will eventually delegitimize education. It is up to the administrations to walk a tightrope, balancing future job security with the value of human work.
Senior Science and Technology Editor
