For districts hiring and teachers job-hunting this spring, chances are artificial intelligence is helping shape the path from application to classroom.
Even more likely, teachers don鈥檛 know it.
That鈥檚 a problem as the technology becomes ubiquitous in districts. Advocates say the systems can cut hiring time and improve teachers鈥� fit in schools, but they may still pose privacy risks and introduce new forms of bias into the hiring process for job-seekers.
Fifty-three percent of district recruiters now use AI tools, according to a nationally representative survey of 270 recruiters fielded this fall by the EdWeek Research Center.
But only 2% of teachers in a connected survey of more than 700 job-seekers told EdWeek that they have applied in the last 12 months to a district that uses AI in hiring. Taken together, the data points suggest 糖心动漫vlog may not realize the technology is being used or have a clear understanding of how its underlying algorithms work.
The and the , the two national teachers鈥� unions, both have guidance on how AI should be used in education, but those resources don鈥檛 specifically cover how the technology shapes teachers鈥� entry into the workplace. Separately, the AFT has created a national academy to train teachers on other aspects of AI in education.
New teachers鈥� contracts in both ., and , include language seeking to limit ways AI could be used to replace or evaluate teachers, but neither covers its use for hiring them in the first place.
鈥淎I is a time-saving device, but it鈥檚 still an early technology, and there are some problems with it,鈥� said Kathryn Cernera, president of the Ithaca teachers鈥� union, 鈥渨hich is where the human decisionmaking piece has to come in.鈥�
Kyra Wilson, who studies the effects of AI on human decisionmaking at the University of Washington, said it鈥檚 more likely than not that the systems will simply reproduce the bias that humans already possess.
鈥淎I [tools] are not a miracle solution for hiring,鈥� she said
What the landscape of AI-assisted teacher hiring looks like
There are no federal data on how many districts are using AI for hiring. Only about school districts have set policies on AI use, according to a recent RAND Corp. study, and virtually none of them cover how the technology can be used in hiring.
But signs point to education following the larger workforce trend of AI-based hiring. Across industries, nearly a third of recruiters now use the technology, according to the human resources group Criteria Research鈥檚 .
Some of the most commonly used hiring software, including PowerSchool鈥檚 and HireVue, use AI to match or rank candidates, among other tasks. HireVue鈥檚 chief science officer, Mike Hudy, said in a statement that the technology has 鈥渕oved from an 鈥榠nteresting experiment鈥� to 鈥榚ssential infrastructure鈥欌�� in school districts the company serves.
Filling deeper district needs
In the last seven years, Golf Middle School Principal David Norman has grappled with longer hiring windows and a shrinking pool of candidates. His 600-student district in Morton Grove, Ill., has no separate human resources department, leaving it up to Norman and other administrators to recruit, manage applications and interviews, and secure new teachers and staff.
鈥淭he hard-to-fill positions have become harder to fill,鈥� Norman said. 鈥淭he candidate pool is changing, and so the way we approach hiring has to change.鈥�
This year, Norman developed a homegrown AI hiring agent based on Google鈥檚 Gemini model that he hopes can help find teachers and support staff who are a good fit for his tiny two-school district 15 miles north of Chicago.
Golf uses an AI agent鈥攁 program that can do complex tasks autonomously鈥攖o screen candidate resumes and help administrators prepare for interviews. Norman uploads applications stripped of identifiable information, and the hiring agent provides analyses of each candidate鈥檚 strengths and challenges as well as how closely they match district job postings.
The agent鈥檚 ability to limit bias in the hiring process has been 鈥渆ye-opening,鈥� Norman said.
The AI agent looks at the compatibility between a candidate鈥檚 redacted resume and skills prioritized in the job description, which 鈥渞emoves those internal questions of 鈥榮hould we?鈥� or 鈥榮houldn鈥檛 we bring them in?鈥� and allows us to quantify why we would bring this person in to interview,鈥� Norman said.
鈥淲hen I first got to administration, it was like 鈥榦h, you went [to my alma mater], I鈥檓 definitely going to interview you,鈥欌�� he said. 鈥淭hose are conversations that I鈥檓 no longer having because of this process, which just makes me a more objective hiring manager.鈥�
Training and caution needed
However, studies suggest it can be difficult to eliminate AI privacy and bias risks fully in the hiring process.
AI tools don鈥檛 operate in a vacuum鈥攖hey learn from existing data, which can be .
鈥淎n AI-enabled teacher hiring system might be assumed to be more objective than human-based r茅sum茅 scoring,鈥� warned the U.S. Education Department鈥檚 now-defunct office of educational technology in a . 鈥淵et, if the AI system relies on poor quality historical data, it might de-prioritize candidates who could bring both diversity and talent to a school鈥檚 teaching workforce.鈥�
States like and , as well as places like New York City, have passed laws or regulations requiring employers鈥攊ncluding school districts鈥攖o ensure equity and privacy protections in their AI-based tools.
In 2023, New York City mandated strict oversight of AI hiring tools, requiring all employers (including school districts) to conduct annual independent bias audits of the tools, post the results, and notify job-seekers within at least 10 business days of their intent to use the tools, with fines up to $1,500 for violations. The law has , mostly because it can be difficult to tell whether a particular software uses AI, or how an underlying AI algorithm makes decisions about job candidates.
Similarly, the algorithms that undergird AI analyze such wide-ranging data that they may be able to identify candidates even when their applications have been stripped of common identifiers. An algorithm might categorize a job-seeker as an older woman of color by patterns of colleges or membership in professional groups, gaps in job history associated with caregiving, and graduation or certification dates.
There have already been high-profile tool and alleging that, among other things, AI systems undervalued job-seekers who attended historically Black or women鈥檚 colleges or deaf applicants whose voices didn鈥檛 match typical speech patterns. Other lawsuits allege the tools by culling personal data from beyond the application without notice.
And while the Golf district and others leave the final decision to administrators, not algorithms, about whom to interview, emerging research suggests people may be more influenced by AI recommendations than they realize.
In their research, Wilson and her colleagues at the University of Washington found when AI hiring tools favor a specific group, 90% of the time鈥攅ven when the people considered the AI鈥檚 recommendations irrelevant or low-quality.
鈥淩ight now AI is still in its infancy in workplace integration, so a lot of people don鈥檛 have the knowledge and don鈥檛 necessarily expect to see these [algorithmic] biases,鈥� Wilson said. District hiring staff need more training on how to evaluate and use AI recommendations, she said.
How to inform human decisions without scaring people off
Teach Away, a for-profit international teacher-recruitment firm that places teachers across 100 countries including the United States, launched its own AI-based system six months ago. AI has helped speed up teacher-school matching on the platform and allows administrators to review information that may get missed on a typical application, such as teachers鈥� experience with particular curricula or student groups, according to CEO David Frey.
But Frey said noticeably automated hiring systems, like chatbots similar to the annoying customer service ones most people sigh over now, can also 鈥渟care off鈥� teachers if they make the hiring process seem cold or disconnected from education.
In a , 42% of international job-hunting teachers on the Teach Away platform said they had spoken to an automated recruiter or AI chatbot as part of the hiring process, and 30% reported withdrawing from a job application because they found the process 鈥渢oo impersonal.鈥�
That hasn鈥檛 been an issue yet for the Golf district鈥檚 hiring agent. Norman, the principal, said it has helped his staff have deeper and more substantive conversations with the teachers they look to hire.
Candidates for a science teacher post, for example, might focus heavily on their experience coaching science instruction or developing curriculum, but few specifically mention differentiation in their resumes or application forms. The hiring agent developed an interview question asking teachers to describe a specific science unit where they successfully differentiated instruction for English learners and students with disabilities.
鈥淎I will not replace the human element in this,鈥� Norman said, 鈥渂ut I think people have to recognize the potential of AI to analyze data and give us ideas.鈥�
And no AI system can make up for the power of recruiters helping potential teachers understand the culture of the school, the types of students, and the challenges new teachers there will face.
鈥淭eachers often want to know they are going to a strong community,鈥� said Frey, 鈥渁nd that鈥檚 harder for teachers to figure out if they鈥檙e just talking to chatbots.鈥�
