Chicago robot tutors are teaching SEL effectively–without pretending to be human

n a crowded fourth-grade classroom in Chicago, a new kind of tutor is shaping how children learn about empathy, conflict, and problem-solving. These robots aren’t programmed to act like friendly classmates with invented emotions and backstories. Instead, they speak plainly, without pretense or fiction, and the results will catch educators’ attention across the country.

The research behind it, led by PhD student Lauren Wright and overseen by Assistant Professor Sarah Sebo at the University of Chicago’s Department of Computer Science, came together thanks to robust partnerships: Chicago Public Schools (CPS) provided access to classrooms and teachers, and Kiljoong Kim at Chapin Hall built crucial connections that made this cross-institution project possible.

“With this work, we wanted to create a team that would be able to uniquely design and study technology, informed by best practices in SEL education, with the input of principals, administrators, teachers, and students in Chicago Public Schools,” said Sebo. “We started not with a specific robot prototype, but by observing SEL instruction in CPS classrooms and talking with teachers about their experiences with SEL, and THEN starting to think about how robots might be able to supplement the amazing work teachers are already doing in schools.“

This research was presented at the 2026 ACM/IEEE International Conference on Human-Robot Interaction (HRI) held in Edinburgh, Scotland, the top venue for human-robot interaction research, where the team not only shared their findings with the broader academic community, but also learned they had won the Best Paper Award—an honor that recognizes the most impactful and innovative research at the conference.

For teachers in CPS, SEL lessons usually mean whole-class activities delivered once a week. In practice, however, many students tune out, and overstretched teachers would love more one-on-one opportunities.Teachers interviewed in the study expressed concern that group SEL lessons rarely reach every child. This perspective, along with careful classroom observation and interviews, drove the research team to look for solutions.

Wright’s team spotted the gap and asked whether robots could supplement teachers and provide individualized instruction where group lessons fall short. Did it actually matter if those robots ‘acted’ human?

In the experiment, fifty-two students participated. One group learned SEL from robots with fictional, emotion-laden dialogue. Another worked with robots that spoke only in factual terms, openly acknowledging they had no feelings or friends. The third group received their regular curriculum with no robot involvement. By leveraging Second Step’s curriculum, made possible through connections with Chapin Hall and Kiljoong Kim, the research team ensured the lessons reflected real classroom needs. Robots adapted group lesson plans into personalized conversations, while teachers continued focusing on the rest of their students.

Both robot groups showed students improved in their mastery of SEL concepts compared to peers who only had classroom instruction. Yet, the factual robots, in their straightforward honesty, often encouraged deeper engagement with lesson vocabulary and problem-solving language. These findings challenge conventional wisdom.

“Giving robots fictional personalities with the intent to make them more engaging is a common approach to educational robots, one which feels especially relevant for teaching SEL,” expressed Wright. “However, in our research study, we found that the robot’s fictional emotions and experiences may have distracted or made students feel less comfortable using lesson language. These findings challenge us to reconsider our assumptions when designing robot behaviors – just because an approach is common doesn’t mean it will always lead to the best outcomes.“

Honest Robots, Authentic Impact

As society becomes more concerned about children forming unhealthy attachments to AI, the Chicago team’s results provide timely guidance. Demonstrating that factual robots can perform as well or better without mimicking emotions points the way to a safer classroom technology.

The central message of the study is clear. Robots are powerful supplements, extending teachers’ capabilities and freeing up attention for students who need more support. But they do not replace the human element in teaching.

“We firmly believe that human teachers are the most important element in elementary education,” said Sebo. “As we all experienced during the pandemic, replacing in-person educational experiences with technology-mediated ones can be disastrous. Our work does not seek to replace human teachers, but instead, aims to create robot tools that extend a teacher’s reach, giving the ability to provide children with one-on-one attention without pulling them away from the rest of the class.”

As this school year winds down, Chicago’s classroom experiment stands as proof of what partnership-driven innovation can achieve in education. The findings invite other districts to rethink how technology can responsibly supplement teachers and ensure every child receives meaningful, individualized support.

---

DOI

10.1145/3757279.3785596 

 

Full-Service Community Schools grantee and partner school characteristics

Today, the Institute of Education Sciences released Initial Challenges Implementing the Full-Service Community Schools Approach and How FY 2023 Grantees Addressed Them (NCEE 2026–004). The Full-Service Community Schools (FSCS) program awards five-year grants to states, districts, or other organizations to help partner schools implement community schools. Community schools are intended to serve as hubs that coordinate and integrate a wide range of services and supports that meet the unique needs of students and communities. This snapshot examines the challenges grantees and their partner schools reported facing in the first year of implementing their grants. Data for this study snapshot comes from the Common Core of Data and surveys of fiscal year 2023 FSCS grantees and their partner schools. Grant awards were made in November 2023, and surveys were administered from May through October 2024.  Findings include: During the first year of the grant, some FY 2023 FSCS grantees and their partner schools reported that it was very challenging to engage families in decision-making activities and to combine funding streams for the program.  Few grantees and partner schools found it challenging to onboard community school coordinators or build relationships with partners. Most partner schools sought support in implementing the community school approach; fewer schools reported receiving technical assistance from grantees, though that assistance addressed a range of topics. To access the study snapshot, please visit: https://ies.ed.gov/use-work/resource-library/report/snapshot/initial-challenges-implementing-full-service-community-schools-approach-and-how-fy-2023-grantees.

Standardized Test Scores and Academic Performance at a Public University System

 Recent studies of Ivy-Plus institutions suggest that standardized test scores (SAT/ACT) are far better predictors of college success than high school grade point average (HS-GPA), prompting a return to the requirement that test scores be submitted for admission at elite colleges. 

This study asks whether re-establishing the SAT requirement for admission at a large urban public university system would improve the predictability of academic outcomes. Using administrative data for the 2010-2019 first-year cohorts, the study updates earlier work of students from public universities as to the relative predictive power of HSGPA and SAT scores on first-year outcomes and graduation rates. 

Contrary to findings at elite private institutions, the study finds that HSGPA is the dominant predictor of academic success in this public system. A one-standard-deviation increase in HSGPA is four to six times more predictive of six-year graduation than a comparable increase in SAT scores. 

Out-of-sample forecasts for the post-COVID period (2020–2024) confirm that test-optional models relying only on HSGPA experience relatively little loss in predictive accuracy compared to models that include test scores. 

HSGPA remains the most reliable signal of degree completion at broad-access public universities.

Cannabis use disorder among young people linked to diagnosis of psychiatric disorders

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

A new study led by Johns Hopkins researchers found that young people with cannabis use disorder were more likely than young people with other substance use disorders to later be diagnosed with a psychiatric disorder. In contrast, adults with cannabis use disorder were significantly less likely to develop psychiatric disorders, compared to adults with other substance use disorders.

The study found that the relative risk of young people age 17 and under with cannabis use disorder was 52% higher for schizophrenia, 30% higher for recurrent major depression, and 21% higher for anxiety disorders, compared to young people with other substance use disorders. Adults with cannabis use disorder had lower relative risks for being diagnosed with a psychiatric disorder.    

The findings highlight the question of whether excessive cannabis use, perhaps more than other substances, might alter brain development of young people, predisposing them to developing a psychiatric disorder.

The study was published online March 5 in the American Journal of Psychiatry.

“Is cannabis use a unique risk factor compared to the use of other substances such as alcohol, opioids, or cocaine? That’s the question we addressed in this study, and our findings suggest that that relative risk depends on the user’s age,” says study co-author Johannes Thrul, PhD, associate professor in the Department of Mental Health at the Bloomberg School.

Products made from the Cannabis sativa plant have been used recreationally in the U.S. since at least the 1800s. Their popularity increased during alcohol prohibition in the 1920s and the counterculture movement in the 1960s. Today, cannabis use by adults age 21 and older is legal in 24 U.S. states and the District of Columbia. Surveys suggest daily use of cannabis is higher than alcohol consumption.

“Much of our interest in this came from the recent legalization of recreational cannabis in Maryland, in 2023, and other states,” says Ryan Nicholson, MD, resident at Johns Hopkins University School of Medicine. “We wanted to understand cannabis-related psychotic disorders clinicians are seeing in the context of other substance-related psychotic disorders."

The link between cannabis and psychosis first appeared in medical literature in the early twentieth century. In 1987, a study of more than 45,000 Swedish army recruits found that the use of cannabis at the time of conscription was associated with large increases in the risk of subsequent schizophrenia, especially when the cannabis use was heavy. Other studies since then have found similar associations.

For the study, the researchers analyzed nearly 700,000 U.S. medical records from a large commercial database. They identified patients who had been diagnosed with cannabis use disorder—a condition that implies relatively heavy cannabis use—but had not been diagnosed with other psychiatric disorders. They then matched these patients on measures such as age, sex, ethnicity, and income level, with patients who had been diagnosed with other, non-cannabis substance use disorders and did not have other psychiatric conditions. The researchers compared the rates of subsequent schizophrenia and other psychiatric diagnoses in these two patient groups—adults age 18 and older (691,806 patients) and one for those age 17 and under (49,586 patients).

The median age among patients with cannabis use disorder was 16 versus 15 among patients with other substance use disorders. About 10% of patients in the cohort of all substance use disorders were under age 12. The authors note that this aligns with reports from adult patients being treated for substance use disorder: 10.2% reported starting substance use at age 11 or younger.

Adults in the cannabis use disorder group had a 19% lower risk (0.34% vs. 0.42%) of subsequent schizophrenia compared to the group with other substance use disorders. Risks of subsequent psychosis, recurrent major depression, and suicide attempts were also lower in the cannabis-use group.

The results are consistent with the idea that heavy cannabis use predisposes young people to subsequent schizophrenia and some other psychiatric disorders that they might not develop otherwise. Thrul notes that this acceleration effect could make these illnesses seem less likely at later ages, thus appearing to lower the risk in adults, at least in relation to other recreational drugs.

Thrul cautions, however, that the causation might point in the other direction, with individuals who are innately more likely to develop certain psychiatric disorders to also have a greater tendency to self-medicate with cannabis, even before their mental health issues have become evident.

“There are still many unknowns on that question, but I would never recommend that teenagers use cannabis, especially not the high-potency cannabis products that are on the market now,” he says.

One of the paper’s limitations is that the database the researchers used relied on International Classification of Diseases ICD-10 coding by other physicians, so the researchers may not know the exact patient history that led to the diagnosis. 

“Association of Cannabis Versus Other Substance Use Disorders with Psychiatric Conditions: A Propensity-Matched Retrospective Cohort Analysis” was co-authored by Ryan Nicholson, Una Choi, Ramin Mojtabai and Johannes Thrul.

The autonomy of universities in the USA falls far behind peer group

 

New report on academic freedom worldwide published

The State of Academic Freedom 2025

Academic freedom continues to decline worldwide. Over the last decade, it has declined in 50 countries, while only 9 countries have registered improvements. The countries that have experienced a decline include several democracies, such as the United States of America (USA), Greece, Finland, and Argentina. Globally, the most widespread declines are in individual academic freedom and in campus integrity. By contrast, fewer countries are experiencing declines in institutional autonomy. However, there are strong reasons to be concerned about attacks on universities’ institutional autonomy.

Institutional Autonomy in a Global Context and in the West

Between 2015 and 2025, institutional autonomy declined in 43 countries, 21 of which are primarily located in Europe, North America, and Latin America. This trend signals a concerning erosion of university autonomy within liberal democracies, for example in Canada, the Netherlands, Switzerland, and the USA, albeit to very different degrees.

Institutional autonomy plays a central role in protecting academic freedom. Although some scholars and political actors have recently argued that strong institutional autonomy might foster universities that suppress viewpoint diversity, the AFI data indicate a positive correlation between institutional autonomy and the freedom to research and teach. Higher education institutions with greater autonomy tend to provide more robust safeguards for academic freedom at the individual level. Conversely, when institutional autonomy is weakened, both universities and individual scholars become more vulnerable to non-academic pressures.

The Case of the United States of America

This year’s AFI Update presents data on the trajectory of institutional autonomy in the USA and compares the data with the regional averages for Western Europe and North America between 2015 and 2025. The analysis shows that the autonomy of universities in the USA has fallen far behind their peer group.

The report also contains data comparing the experience in the USA with other formerly democratic countries undergoing autocratization, such as India, Hungary, and Türkiye. In these cases, after previously high levels of academic freedom, declines in institutional autonomy have occurred over longer periods and with varying magnitudes. By contrast, the decline in the USA has been comparatively abrupt, with institutional autonomy falling from 2.4 in 2024 to 1.7 in 2025 (on a scale from 0-4). This significant drop in just one year was driven mostly by coercive federal actions. The decline first started in 2020 due to state-level measures that have continued to put universities under pressure. In 2019, the score for institutional autonomy in the USA was still 3.3, a comparatively high score in both global and regional comparisons.

Despite the alarming erosion of university autonomy in the USA, resistance has emerged through legal and institutional pushback. Actions by judicial bodies, academic actors, and civil society organizations suggest that avenues to mitigate or reverse this harmful development may still remain.

Data

This year’s Academic Freedom Index Update is based on data from V-Dem’s version 16 release, which draws on assessments made by 2,357 country experts from around the world. The data cover the period from 1900 to 2025. All data are publicly available and include a total of more than one million data points at the coder level. The aggregate index is composed of five indicators, namely the freedom to research and teach; the freedom of academic exchange and dissemination; the institutional autonomy of universities; campus integrity; and the freedom of academic and cultural expression.

Open Access und Visualisierungen

The data used for the 2026 AFI Update are available in open access to facilitate further studies. Please also visit the Academic Freedom Index website, where you will find the full report, an interactive visualization of the data, country profiles, and information on the index project. The German Volkswagen Foundation has provided funding for the project for a total of five years starting in 2021.

Additional easy-to-use graphing tools are also available on the V-Dem website for anyone who is interested; they can be consulted by researchers, students, university administrators, research funders, and policy-makers.

---

Whose turn is it to speak? Toddlers know the answer!

 

Two‑year‑olds already have a surprisingly good sense of when it is someone’s turn to speak. This is shown by new research conducted by linguist Imme Lammertink from the Baby & Child Research Centre at Radboud University, together with colleagues from the Max Planck Institute and the University of Chicago. That’s impressive, because it requires detailed knowledge of language and sentence structure - skills we know children do not fully possess at this age.

As adults engaged in conversation, we constantly predict when the other person will finish speaking and whether a response is expected from us. This helps us avoid long, awkward silences or talking over someone. New research by Lammertink and colleagues shows that children from the age of two already use linguistic cues to predict whose turn it is to speak. “Apparently, even without fully understanding sentences, young children are able to anticipate when a turn switch will occur,” Lammertink explains.

Is It Your Turn?

In interrogative sentences, the verb often appears at the beginning ( “may you put them outside?”), and the pronoun “you” is used more frequently than “I” (“shall I leave them outside then?”). The researchers showed children an animation of two characters engaged in conversation while tracking the children’s eye movements. “We examined whether children looked towards the other character before the speaking character had finished talking. If they did, this meant they were predicting that a turn switch was about to happen,” Lammertink explains. Toddlers made more predictive eye movements during questions containing “you” than during questions containing “I”. This shows that toddlers already realise that, after questions with “you”, someone else is more likely to speak than after questions with “I”.

The researchers also observed that the number of predictive eye movements increases with age. “One‑year‑olds did not pick up on these signals, but from two years onwards, children become increasingly better at predicting turn‑taking: four‑year‑olds predict even more than two‑ and three‑year‑olds.”

Children with Developmental Language Disorder

The use of these linguistic cues is linked to language ability. Therefore, the researchers also investigated whether children suspected of having a Developmental Language Disorder (DLD) are able to predict turn switches. Lammertink explains: “Children with DLD struggle significantly with language. They often start speaking late, have difficulty understanding others, can be hard to understand themselves, and produce short sentences. But our study showed that three‑year‑olds with DLD, just like their peers without DLD, are able to predict turn‑taking.”

Children with DLD were, however, slower in making their predictions than children without DLD. This may explain why children with DLD sometimes respond more slowly in conversation. The research shows that children with DLD do understand when it is their turn to speak. They also understand that a response is expected after a question. “By asking these children more questions, you help them practise switching turns,” says the linguist. “It helps to begin these questions with a verb and to use the personal pronoun ‘you’. This enables them to pick up more quickly on the signal that a response is expected from them.”

---

Journal

Journal of Speech Language and Hearing Research

DOI

10.1044/2025_JSLHR-25-00180 

Link between ADHD and mental ill health in teens

 Scientists have shed light on some reasons why young people with attention deficit hyperactivity disorder – known as ADHD – are prone to anxiety and depression.

Young people with ADHD symptoms are more likely to experience emotional issues such as anxiety and low mood partly because ADHD puts them at risk of low self-esteem and having a parent with poor mental health.

The study, led by the University of Edinburgh and funded by the Medical Research Foundation, is the first to study a wide range of factors which link ADHD symptoms and mental ill health symptoms over time.

It is estimated that ADHD affects around five per cent of children and young people in the UK. One in four young people with ADHD have an anxiety disorder and 40 per cent experience depressive episodes.

Researchers from the University of Edinburgh examined survey data from over 5,000 adolescents aged 11 to 17 from the UK Millennium Cohort Study, which follows young people born between 2000 and 2002 across the UK.

They tested fourteen possible factors linking ADHD and mental ill health, including relationships with family and friends, behaviour issues and low self-esteem.  

Other factors assessed were behaviour at school, general health and if their parent had mental health difficulties.

The team analysed questionnaires which surveyed parents and the young people on ADHD symptoms and emotional problems such as low mood and anxiety symptoms at ages 11,14 and 17.

Findings showed that self-esteem and parental mental health had a small but statistically significant link to both ADHD and mental ill health risk. Among girls difficulties with peers had a small but significant link.

This was the case even when other factors such as any pre-existing neurodevelopmental and mental health issues were taken into account.

The results suggest that several different factors may be working collectively with a small effect to connect ADHD and other emotional issues during adolescence. 

The study could help in tailoring targeted wellbeing support systems for young people with ADHD, experts say.

Professor Aja Murray, of the University of Edinburgh’s School of Philosophy, Psychology and Language Sciences, who led the study, said: “The findings suggest that to help reduce the risk of adolescents with ADHD symptoms developing other mental ill health symptoms, two things are supporting parents to improve their mental health, and supporting adolescents to foster high self-esteem. For girls in particular supporting social skills development is also important.”

Dr Angela Hind, Chief Executive of the Medical Research Foundation, said:

“ADHD can have a profound impact on a young person’s life, yet relatively little is known about the mental health challenges they might face as they grow up. This important research sheds new light on why teenagers with ADHD are at greater risk of anxiety and depression than those without, highlighting self-esteem and a parent’s mental health as two of the most crucial factors that shape their wellbeing.

"These findings demonstrate the importance of funding medical research into children and young people’s mental health, and bring us a step closer to developing more targeted support for teenagers with ADHD, ensuring they can thrive during some of their most formative years.”

The study is published in the Journal of Attention Disorders, link to study:  https://pubmed.ncbi.nlm.nih.gov/41789525/