The robots have 

New research suggests that a majority of  at school. To write essays. To solve complicated math problems. To find love. 

Wait, what? 

Nearly a fifth of students said they or a friend have used artificial intelligence chatbots to form romantic relationships, according to . Some 42% said they or someone they know used the chatbots for mental health support, as an escape from real life or as a friend.

Eighty-six percent of students say they鈥檝e used artificial intelligence chatbots in the past academic year 鈥� half to help with schoolwork.

The tech-enabled convenience, researchers conclude, doesn’t come without significant risks for young people. Namely, as AI proliferates in schools 鈥� with help from the federal government and a zealous tech industry 鈥� on a promise to improve student outcomes, they warn that young people could grow socially and emotionally disconnected from the humans in their lives. 

• Dig Deeper: 

In the news

The latest in Trump鈥檚 immigration crackdown: The survey featured above, which quizzed students, teachers and parents, also offers startling findings on immigration enforcement in schools: 
While more than a quarter of educators said their school collects information about whether a student is undocumented, 17% said their district shares records 鈥� including grades and disciplinary information 鈥� with immigration enforcement. 

In the last school year, 13% of teachers said a staff member at their school reported a student or parent to immigration enforcement of their own accord. | 

• Call for answers: In the wake of immigration enforcement that鈥檚 ensnared children, New York congressional Democrats are demanding the feds release information about the welfare of students held in detention, my colleague Jo Napolitano reports. | 
• A 13-year-old boy from Brazil, who has lived in a Boston suburb since 2021 with a pending asylum application, was scooped up by Immigration and Customs Enforcement after local police arrested him on a 鈥渃redible tip鈥� accusing him of making 鈥渁 violent threat鈥� against a classmate at school. The boy鈥檚 mother said her son wound up in a Virginia detention facility and was 鈥渄esperate, saying ICE had taken him.鈥� | 
• Chicago teenagers are among a group of activists patrolling the city鈥檚 neighborhoods to monitor ICE鈥檚 deployment to the city and help migrants avoid arrest. | 
• Immigration agents detained a Chicago Public Schools vendor employee outside a school, prompting educators to move physical education classes indoors out of an 鈥渁bundance of caution.鈥� | 
• A Des Moines, Iowa, high schooler was detained by ICE during a routine immigration check-in, placed in a Louisiana detention center and deported to Central America fewer than two weeks later. |
• A 15-year-old boy with disabilities 鈥� who was handcuffed outside a Los Angeles high school after immigration agents mistook him for a suspect 鈥� is among more than 170 U.S. citizens, including nearly 20 children, who have been detained during the first nine months of the president’s immigration push. | 
  

Trigger warning: After a Washington state teenager hanged himself on camera, the 13-year-old boy鈥檚 parents set out to find out what motivated their child to livestream his suicide on Instagram while online users watched. Evidence pointed to a sadistic online group that relies on torment, blackmail and coercion to weed out teens they deem weak. | 

Civil rights advocates in New York are sounding the alarm over a Long Island school district鈥檚 new AI-powered surveillance system, which includes round-the-clock audio monitoring with in-classroom microphones. | 

A federal judge has ordered the Department of Defense to restock hundreds of books after a lawsuit alleged students were banned from checking out texts related to race and gender from school libraries on military bases in violation of the First Amendment. | 

More than 600 armed volunteers in Utah have been approved to patrol campuses across the state to comply with a new law requiring armed security. Called school guardians, the volunteers are existing school employees who agree to be trained by local law enforcement and carry guns on campus. | 

No 鈥淛ackass鈥�: Instagram announced new PG-13 content features that restrict teenagers from viewing posts that contain sex, drugs and 鈥渞isky stunts.鈥� | 

A Tuscaloosa, Alabama, school resource officer restrained and handcuffed a county commissioner after a spat at an elementary school awards program. | 

The number of guns found at Minnesota schools has increased nearly threefold in the last several years, new state data show. | 

More than half of Florida鈥檚 school districts received bomb threats on a single evening last week. The threats weren鈥檛 credible, officials said, and appeared to be 鈥減art of a hoax intended to solicit money.鈥� | 

ICYMI @The74

Emotional Support

鈥淏ig picture, AI is not very good at math,鈥� said Alex Kotran, co-founder and CEO of . 鈥淟anguage models just predict the next word. You get mixed results using language models to do math. It’s not yet mature enough to where it can be trusted to be scaled.鈥�

And even if it were to improve, critics worry it might hurt kids鈥� ability to try 鈥� and fail 鈥� on their own. Much would be lost, Kotran said, if 鈥渨e get rid of productive struggle and we build this instinct where the first thing you do is go to AI for help.鈥�

But students in the United States and the United Kingdom have a different view. found 56% believe AI could go a long way in reducing math anxiety. 

Fifteen percent of the 1,500 16- to 18-year-old students surveyed said they had already experienced this relief themselves and slightly more than 1 in 5 said their math scores improved because of the technology. 

The survey also included . Sixty-one percent suggested students view AI as 鈥渁 mentor or study partner rather than a crutch鈥�, while nearly half 鈥渟ee value for students in using AI for help with the process of learning math concepts, rather than to give answers.鈥� 

Nicole Paxton, principal of Mountain Vista Community School in Colorado Springs, said her teachers use AI in many ways. Tools like MagicSchoolAI analyze student responses to math prompts, with AI generating 鈥渟pecific, standards-aligned feedback for each student, focusing on their reasoning, accuracy, and math vocabulary.鈥� 

Paxton said the tool highlights strengths and misconceptions, 鈥渨hich helps teachers give timely and targeted next steps.鈥� The practice saves educators time so they can 鈥渕ore easily differentiate their re-teaching or follow-up, especially when addressing common errors across the class.鈥�

Robin Lake, director of the Center on Reinventing Public Education, recently looked at the evidence base for using AI in math instruction, including whether it can help the “shocking number of students” with foundational skills’ gaps like those identified in a recent study. 

The May 13 analysis by TNTP found that almost half of the students sampled started the class with only one-third of the concepts and skills needed from earlier grades. Lake said AI can be used by schools to identify children who are struggling 鈥� and, at least to some degree, by the students themselves. 

鈥淎I can be very helpful in analyzing data and identifying gaps in student learning,鈥� she said.

And, if a student wants to learn a mathematical concept in a different way than what they鈥檝e experienced in class, she said, AI can provide a valuable alternative. 

鈥淎 lot of students are already doing this,鈥� Lake said. 

to use the technology, though many educators Terrie Galanti, associate professor at the University of North Florida, said AI success in student learning depends on how teachers are prepared to use it. 

鈥淎I can be more than an explainer or an answer giver,鈥� said Galanti, who teaches secondary mathematics and STEM integration/computational thinking. 鈥淲ith thoughtful prompts, AI can become part of interactive, collaborative conversations to deepen mathematics understanding.鈥� 

鈥嬧�婽he National Council of Teachers of Mathematics said in that teachers have long been accustomed to technological advances that change the way students learn. 

They had already adjusted to the availability of pocket calculators in the early 鈥�80s and, more recently, to the widespread use of , a mobile app that recognizes and solves math problems. 

It notes that advancements in AI make teachers more, not less valuable, in student learning. 

Latrenda Knighten, the organization鈥檚 president, told in March that students will still need to rely upon their own discernment to solve mathematical problems 鈥� regardless of what tools become available.  

鈥淲e know that children learn math from being able to problem-solve, being able to use reasoning skills, critical thinking, having opportunities to collaborate with each other and talk about what they鈥檙e doing,鈥� Knighten said. 

Irina Lyublinskaya, professor in the department of mathematics, science, and technology at Teachers College, Columbia University, distinguished between chatbots like ChatGPT and computational knowledge engines like She noted math specific AI-powered applications 鈥� including WolframAlpha and Symbolab 鈥� work very well. 

鈥淎I chatbots can help students learn math, and they can help teachers to support students, but this is not about asking ChatGPT to solve a math problem,鈥� she said. 鈥淚 know of research-based initiatives that use AI to adapt learning materials to students’ learning styles and abilities and these definitely help students learn.鈥�

One, she noted, was , developed by researchers and educators in Europe, and is now being tested in NYC. 

鈥淐hatbots can be trained as teaching assistants or tutors that can provide students proper scaffolding and feedback, helping them to learn math the same way they would with a real person,鈥� she said. 

Zachary A. Pardos is an associate professor of education at the University of California Berkeley where he studies adaptive learning and AI. He found, in conducted a year ago, that 25% of the answers provided by ChatGPT in algebra were incorrect. 

鈥淭hat鈥檚 pretty high,鈥� he noted. 鈥淢uch higher than you would want.鈥�

But the technology has improved since then. 

鈥淲ith the right techniques 鈥� at least in algebra 鈥� from an error perspective, I feel it is ready for real-time intervention in math,鈥� he said.

The AI bot, created by Boston-based AllHere Education, was launched . But just three months later, AllHere posted that a majority of its 50 or so employees had been furloughed due to its 鈥渃urrent financial position.鈥� A spokesperson for the Los Angeles district said company founder and CEO Joanna Smith-Griffin was no longer on the job. AllHere was up for sale, the district said, with several businesses interested in acquiring it.

The news was shocking and certainly bleak for the ed tech industry, but several observers say the partnership bit off more than it could chew, tech-wise 鈥� and that the ensuing blowup could hurt future AI investments.

Ed was touted as a powerful, easy-to-use o for students and parents to supplement classroom instruction, find assistance with kids鈥� academic struggles and help families navigate attendance, grades, transportation and other key issues, all in 100 languages and on their mobile phones.

But Amanda Bickerstaff, founder and CEO of , a consulting and training firm, said that was an overreach.

鈥淲hat they were trying to do is really not possible with where the technology is today,鈥� she said. 鈥滻t’s a very broad application [with] multiple users 鈥� teachers, students, leaders and family members 鈥� and it pulled in data from multiple systems.鈥�

She noted that even a mega-corporation like McDonald鈥檚 had to trim its AI sails. The fast-food giant recently admitted that a small experiment using a chatbot to power drive-thru windows had resulted in a few fraught customer interactions, such as one in which a woman angrily tried to persuade the bot that she wanted a caramel ice cream as it added to her order.

If McDonald’s, worth an estimated $178.6 billion, can鈥檛 get 100 drive-thrus to take lunch orders with generative AI, she said, the tech isn鈥檛 鈥渨here we need it to be.鈥�

If anything, L.A. and AllHere did not seem worried about the project鈥檚 scale, even if industry insiders now say it was bound to under-deliver: Last spring, at a series of high-profile ed tech conferences, Smith-Griffin and Superintendent Alberto Carvalho showed off Ed widely, with Carvalho saying it would revolutionize students鈥� and parents鈥� relationships to school, 鈥渦tilizing the data-rich environment that we have for every kid.鈥�

In an interview with 社区黑料 at the ASU+GSV Summit in San Diego in April, Carvalho said many students are not connected to school, 鈥渢herefore they’re lost.鈥� Ed, he promised, would change that, with a 鈥渟ignificantly different approach鈥� to communication from the district.

鈥淲e are shifting from a system of 540,000 students into 540,000 鈥榮chools of one,鈥欌�� with personalization and individualization for each student, he said, and 鈥渕eaningful connections with parents.鈥�

Better communication with parents, he said, would help improve not just attendance but reading and math proficiency, graduation rates and other outcomes. 鈥淭he question that needs to be asked is: Why have those resources not meaningfully connected with students and parents, and why have they not resulted in this explosive experience in terms of educational opportunity?鈥�

Carvalho noted Ed鈥檚 ability to understand and communicate in about 100 different languages. And, he crowed, it 鈥渘ever goes to sleep鈥� so it can answer questions 24/7. He called it 鈥渁n entity that learns and relearns all the time and does nothing more, nothing less than adapt itself to you. I think that’s a game changer.鈥� 

But one experienced ed tech insider recalled hearing Carvalho at the conference in April and say it was already solving 鈥渁ll the problems鈥� that big districts face. The insider, who asked not to be identified in order to speak freely about sensitive matters, found the remarks troubling. 鈥淭he messaging was so wrong that at that point I basically started a stopwatch on how long it would take鈥� for the effort to fail. 鈥淎nd I’m kind of amazed it’s been this long before it all fell apart. I feel badly about it, I really do, but it’s not a surprise.鈥�

鈥楢 high-risk proposition鈥�

In addition to the deal鈥檚 dissolution, 社区黑料 reported last week that a former senior director of software engineering at AllHere told district officials, L.A.鈥檚 independent inspector general鈥檚 office and state education officials that Ed processed student records in ways that likely ran afoul of the district鈥檚 own data privacy rules and put sensitive information at risk of being hacked 鈥� warnings that he said the agencies ignored. 

AI for Education鈥檚 Bickerstaff said developers 鈥渉ave to take caution鈥� when building these systems for schools, especially those like Ed that bring together such large sets of data under one application.

鈥淭hese tools, we don’t know how they work directly,鈥� she said. 鈥淲e know they have bias. And we know they’re not reliable. We know they can be leaky. And so we have to be really careful, especially with kids that have protected data.鈥�

Alex Spurrier, an associate partner with the education consulting firm , said what often happens is that district leaders 鈥渢ry to go really big and move really fast to adopt a new technology,鈥� not fully appreciating that it鈥檚 鈥渁 really high risk proposition.鈥�

While ed tech is of overpromising and disappointing results, Spurrier said, other districts dare to take a different approach, starting small, iterating and scaling up. In those cases, he said, disaster rarely follows.

Richard Culatta, CEO of the (ISTE), put it more bluntly: 鈥淲henever a district says, ‘Our strategy around AI is to buy a tool,’ that’s a problem. When the district says, ‘For us, AI is a variety of tools and skills that we are working on together,’ that’s when I feel comfortable that we’re moving in the right direction.鈥�

Culatta suggested that since generative AI is developing and changing so rapidly, districts should use the next few months as 鈥渁 moment of exploration 鈥� it’s a moment to bring in teachers and parents and students to give feedback,鈥� he said. 鈥淚t is not the moment for ribbon cutting.鈥� 

鈥業t’s about exploring鈥�

Smith-Griffin founded AllHere in 2016 at Harvard University鈥檚 . In an April interview with 社区黑料, she said she originally envisioned it as a way to help school systems reduce chronic absenteeism through better communication with parents. Many interventions that schools rely on, such as phone calls, postcards and home visits, 鈥渢end to be heavily reliant on the sheer power of educators to solve system-wide issues,鈥� she said.

A former middle-school math teacher, Smith-Griffin recalled, 鈥淚 was one of those teachers who was doing phone calls, leaving voicemails, visiting my parents’ homes.鈥� 

AllHere pioneered text messaging 鈥渘udges,鈥� electronic versions of postcard reminders to families that, in one key study, modestly. 

The company鈥檚 for L.A., Smith-Griffin said, envisioned extending the attendance strategies while applying them to student learning 鈥渋n the most disciplined way possible.鈥�

鈥淵ou nudge a parent around absences and they will tell you things ranging from, 鈥楳y kid needs tutoring, my kid is struggling with math鈥� [to] 鈥業 struggle with reading,鈥欌�� she said. AllHere went one step further, she said, bringing together 鈥渢he full body of resources鈥� that a school system can offer parents.

The district had high hopes for the chatbot, requiring it to focus on 鈥渆liminating opportunity gaps, promoting whole-child well-being, building stronger relationships with students and families, and providing accessible information,鈥� according to the proposal.

In April, it was still in early implementation at 100 of the district鈥檚 lowest performing 鈥減riority鈥� schools, serving about 55,000 students. LAUSD planned to roll out Ed for all families this fall. The district 鈥渦nplugged鈥� the chatbot on June 14, the Los Angeles Times , but a district spokesperson said L.A. 鈥渨ill continue making Ed available as a tool to its students and families and is closely monitoring the potential acquisition of AllHere.鈥� The company did not immediately responded to queries about the chatbot or its future.

As for the apparent collapse of AllHere, speculation in the ed tech world is rampant.

In the , education entrepreneur Ben Kornell said late last month, 鈥淢y spidey sense basically goes to ‘Something’s not adding up here and there’s more to the story.’鈥� He theorized a 鈥渃ritical failure point鈥� that鈥檚 yet to emerge 鈥渂ecause you don’t see things like this fall apart this quickly, this immediately鈥� for such a small company, especially in the middle of a $6 million contract.

Kornell said the possibilities fall into just a few categories: an accounting or financial misstep, a breakdown among AllHere鈥檚 staff, board and funders or 鈥渕ajor customer payment issues.鈥� 

The district also may have withheld payment for undelivered products, but he said the sudden collapse of the company seemed unusual. 鈥淚f you are headed towards a cash crisis, the normal thing to do would be: Go to your board, go to your funders, and get a bridge to get you through that period and land the plane.鈥�

Bellwether鈥檚 Spurrier said L.A. deserves a measure of credit 鈥渇or being willing to lean into AI technology and think about ways that it could work.鈥� But he wonders whether the best use of generative AI at this moment will be found not in 鈥渞evolutionizing instruction,鈥� as L.A. has pursued, but elsewhere. 

鈥淭here’s plenty of opportunities to think about how AI might help on the administrative side of things, or help folks that are kind of outside the classroom walls,鈥� rather than focusing on changing how schools deliver instruction. 鈥淚 think that’s the wrong place to start.鈥�

ISTE鈥檚 Culatta noted that just down the road from Los Angeles, in Santa Ana, California, district officials there responded to the dawn of tools like ChatGPT and Google鈥檚 Gemini by creating evening classes for adults. 鈥淭he parents come in and they talk about what AI is, how they should be thinking about it,鈥� he said. 鈥淚t’s about exploring. It’s about helping people build their skills.鈥� 

鈥楬ow are your financials?鈥�

The fate of AllHere鈥檚 attendance work in districts nationwide isn鈥檛 clear at the moment. In one large district, the Prince George鈥檚 County, Maryland, Public Schools, near Washington, D.C., teachers piloted AllHere with 32 schools as far back as January 2020, spokeswoman Meghan Thornton said. The district added two more schools to the pilot in 2022, but AllHere notified the district on June 18 that, effective immediately, it wouldn鈥檛 be able to continue its services due to 鈥渦nforeseen financial circumstances.鈥� 

District officials are now looking for another messaging system to replace AllHere 鈥渟hould it no longer be available,鈥� Thornton said.

Bickerstaff said the field more broadly suffers from 鈥渁 major, major overestimation of the capabilities of the technology to date.鈥� L.A., she noted, is the nation鈥檚 second-largest school district, so even the pilot stage likely saw 鈥渧ery high鈥� usage, raising its costs. She predicted a fast acquisition of AllHere, noting that they鈥檇 been looking for outside investment for several months.

As founder of the startup , which offers teachers tools to streamline their workload, Adeel Khan is no stranger to hustling for funding 鈥� and to competitors running out of money. But he said the news about AllHere and Ed was bad for the industry more broadly, leaving districts with questions about whether to partner with newer, untested companies.

鈥淚 see it as something that is certainly not great for the startup ecosystem,鈥� he said.

Even before the news about AllHere broke last month, Khan attended ISTE鈥檚 big national conference in Denver last month, where he talked to school district officials about prospective partnerships. 鈥淢ore than one time I was asked directly, ‘How are your financials?’鈥� he recalled. 

Usually technology directors ask about features and what a product can do for students, he said. But they鈥檙e beginning to realize that a failed product doesn鈥檛 just waste time and money. It damages reputations as well. 鈥淭hat is on the mind of buyers,鈥� he said. 

When school districts invest in new tech, he said, they鈥檙e not just committing to funding it for months or even years, but also to training teachers and others, so they want responsible growth.

鈥淭here’s a lot of disruption to K-12 when a product goes out of business,鈥� Khan said. 鈥淪o people remember this. They remember, ‘Hey, we committed to this product. We discovered it at ISTE two years ago and we loved it. It was great 鈥� and it’s not here anymore. And we don’t want to go through that again.鈥� 鈥�

But this fall, a few educators are quietly charting a different course they believe could change everything: At least two groups are pushing to create new AI chatbots that would offer teachers unlimited access to sometimes confusing and often paywalled peer-reviewed research on the topics that most bedevil them. 

Their aspiration is to offer new tools that are more focused and helpful than wide-ranging ones like ChatGPT, which tends to stumble over research questions with competing findings. And like many kids faced with questions they can’t answer, it has a frustrating tendency to make things up.

Tapping into curated research bases and filtering out lousy results would also make the bots more reliable: If all goes according to plans, they鈥檇 cite their sources.

The result, supporters say, could revolutionize education. If their work takes hold, millions of teachers for the first time could routinely access high-quality research and make it part of their everyday workflow. Such tools could also help stamp out adherence to stubborn but ill-supported fads in areas from 鈥渓earning styles鈥� to reading instruction.

So far, the two groups are each feeling their way around the vast undertaking, with slightly different approaches.

In June, the International Society for Technology in Education introduced , a tool built on content vetted by ISTE and the Association for Supervision and Curriculum Development. (The two groups merged in 2022.) ISTE has made it available in to selected users. All of the chatbot鈥檚 content is educator-focused, and it鈥檚 trained solely on materials developed or approved by the two organizations. 

Now its creators say that within about six months, they expect that the tool will also be able to scour outside, peer-reviewed education research and return 鈥減retty understandable, pretty meaningful results鈥� from vetted journals, said Richard Culatta, ISTE鈥檚 CEO.

鈥淭here’s this big gap between what we know in the research and what happens in practice,鈥� he said. One reason: Most research is published in a format that 鈥渋s just totally inaccessible to teachers.鈥�

Case in point: A set of by the Jefferson Education Exchange, a nonprofit supported by the University of Virginia鈥檚 Curry School of Education, found that while educators prefer research they can act on 鈥� and that鈥檚 presented in a way that applies to their work 鈥� only about 16% of teachers actually use research to inform instruction.

So he and others are building a digital tool, 鈥減urpose-built for educators by educators,鈥� that can translate research into practice, using 鈥渧ery practical language that teachers understand.鈥�

For instance, a teacher could ask the chatbot, 鈥淲hat does the research say about creating a healthy school culture?鈥� or 鈥淲hat鈥檚 the evidence for teaching phonics to developing readers?鈥� One could also ask it to suggest activities that are appropriate for middle school students learning about digital citizenship.

Joseph South, ISTE鈥檚 chief learning officer, said teachers want the latest research, but are up against formidable obstacles. 鈥淭hey have to find the article in the journal that happens to relate to the thing that they want to do,鈥� he said. 鈥淭hey have to somehow understand academic-speak. They have to have the time to read this, and they have to translate it into something useful.鈥�

While ChatGPT can comb through journals it has access to, translate and summarize the research, he said, it鈥檚 not reliable. The typical chatbot 鈥� and thus the typical end user 鈥� doesn’t know whether the results are from a credible, peer-reviewed journal or not, and it may not necessarily care.

鈥淲e do, though,鈥� he said. 鈥淪o we can do that filtering and let the AI do its magic.鈥�

As with its beta version, the new chatbot will also cite the sources used to generate each response. And it鈥檒l let users know when it simply doesn鈥檛 have enough information to return a reliable response.

Developers are still in the early stages of deciding what academic journals to include. For now, they鈥檙e experimenting with a handful of key research articles, but will expand the chatbot鈥檚 range if initial prototypes prove helpful to educators.

Culatta and South, both veterans of the U.S. Department of Education, have spent years working on the research-to-practice problem, offering, in effect, translation services for research findings. 鈥淲e’ve spent so much work trying to figure out how to do it and it’s just never really worked,鈥� he said. 鈥淚t’s just always been a struggle. And we actually think that this could be the first for-real, sustainable, scalable approach to taking research and getting it into language that actually could be used by teachers.鈥�

, a professor of psychology at the University of Virginia and a well-known translator of education research, said his limited experience with ChatGPT has shown that when asked about a subject where there鈥檚 general consensus, such as “What is the effect of sleep on memory?” it produces helpful results. But it isn鈥檛 very good at synthesizing conflicting findings.

It鈥檚 also inconsistent in its willingness to reveal, in Willingham鈥檚 words, that 鈥溾�業 really don’t know anything about that.鈥� And so it, you know, just .鈥�

A paid ChatGPT subscriber, Willingham said he gets 鈥渞eally useful鈥� results only about 20% of the time. 鈥淏ut it requires plenty of verification on my part. And this is all within my area of expertise, so it’s not very hard for me to verify.鈥�

Tapping 鈥榃hat Works鈥�

ISTE isn鈥檛 the only organization pushing to make education research more widely accessible via chatbot. The Learning Agency, a Washington, D.C.-based consulting firm, is also testing a of a bot designed to offer answers to education research queries.

Unlike ISTE鈥檚, the agency鈥檚 tool taps an already existing, if finite, resource: the U.S. Department of Education鈥檚 What Works Clearinghouse, or more specifically its , a curated collection of materials developed by the department鈥檚 Institute of Education Sciences.

鈥淲e were inspired to basically create a special version of ChatGPT that was exposed to more high-quality educational data and research evidence on what works,鈥� said Perpetual Baffour, the group鈥檚 research director.

In a sense, she said, much of the work had already been done, since the library, though limited, exists to translate research findings into more digestible forms for educators. The result is a prototype that offers what Baffour calls 鈥渆vidence-based education advice鈥� on topics from adolescent literacy to dropout prevention and school turnarounds.

Baffour noted that the app development tool allowed the agency to create a 鈥渟imple but robust鈥� prototype within a day.

At the moment, the version they鈥檙e testing is 鈥渧ery basic,鈥� Baffour said. The agency is still considering what it鈥檒l look like in the future. 

鈥淚t only knows as much as the Doing What Works Library tells it,鈥� she said. So queries about topics that are in the library return rich results. But asking it about topics that aren鈥檛 can be problematic. 

For example, ask it about myths around the aforementioned learning styles and it defaults to a more general knowledge base scraped from Wikipedia articles, transcripts of recorded conversations and materials from 鈥渄ifferent corners of the Internet,鈥� Baffour said. 鈥淎nd as you can imagine, those sources might not have the most up-to-date and accurate information about education,鈥� Baffour said.

Indeed, a query about the topic returns this: 鈥淭he myth concerning learning styles is that there is one specific style that works best for everyone. This is not true, as different people learn in different ways and have different preferences for how they learn. Additionally, there is no evidence that suggests that focusing on one particular learning style is more effective than focusing on multiple styles.鈥�

Not exactly accurate or helpful.

In the first place, the widely believed 鈥渕yth鈥� holds that people with different learning styles learn best when their preferred mode of learning is indulged 鈥� not that one style works best for everyone. At a more basic level, while many people may express preferences for ways to take in new information and study 鈥� receiving instruction verbally, for example, instead of via pictures 鈥� scientists have yet to find good evidence that material tuned to these preferences . 

Unfortunately, at the moment the agency鈥檚 bot doesn鈥檛 confess whether it knows a lot or little about a topic. Baffour said they want to change that soon. For now, however, that鈥檚 just an aspiration.

鈥淚 think you’re more likely to get a confident chatbot producing inaccurate information than you are to get a self-aware chatbot admitting its false and incomplete knowledge,鈥� she said. 

Willingham, the UVA researcher, said a useful education-focused chatbot would not just have to incorporate reliable findings, but put them in context. For example, an answer to a query about the evidence for phonics instruction would properly note that, while the record is fairly strong, a lot of mediocre research and 鈥渉yperbolic claims鈥� made in support of alternative methods serve to cloud the overall picture 鈥� a delicate but accurate detail.

鈥淗ow is an aggregator going to negotiate that?鈥� he said. 

Asked if he thought a chatbot might soon replace him, Willingham, the author of and a that translate learning science into plain English, said he wouldn鈥檛 make any predictions. 

鈥淚 was never much of a futurist, but I hocked my crystal ball 15 years ago,鈥� he said.

A new generation of sophisticated systems has emerged in the last year, including Open AI’s GPT-4. These so-called large-language models employ neutral networks trained on massive data sets to generate text that is extremely human-like. By understanding context and analyzing patterns, they can produce relevant, coherent and creative responses to prompts. 

Based on my experiences using several of these systems over the past year, I believe that society may be in the early stages of a transformative moment, similar to the introduction of the web browser and the smartphone. These nascent iterations have flaws and limitations, but they provide a glimpse into what might be possible on the very near horizon, where AI assistants liberate educators from mundane and tedious tasks, allowing them to spend more time with students. And this may very well usher in an era of individualized learning, empowering all students to realize their full potential and fostering a more equitable and effective educational experience.

There are four reasons why this generation of AI tools is likely to succeed where other technologies have failed:

1. Smarter capabilities: These AI systems are now capable of , from high school to graduate- and professional-level exams that span . Google’s Med-PaLM performed at an 鈥� on the medical licensing exam, not only correctly answering the questions but also providing a rationale for its responses. The rate of improvement with these systems is astonishing. For example, GPT-4 made significant progress in just four months, going from a failing grade on the bar exam to scoring in the 90th percentile. It scored in the 93rd percentile on the SAT reading and writing test and the 88th on the LSAT, and got a 5 鈥� the top score 鈥� on several Advanced Placement exams.
   
2. Reasoning engines: AI models like GPT-4, Microsoft’s Bing Chat, and Google’s Bard are advancing beyond simple knowledge repositories. They are developing into sophisticated that can contextualize, infer and deduce information in a manner strikingly similar to . While traditional search engines functioned like librarians guiding users toward relevant resources, this new generation of AI tools acts as skilled graduate research assistants. They can be tasked with requests such as conducting literature reviews, analyzing data or text, synthesizing findings and generating content, stories and tailored lesson plans.
   
3. Language is the interface: One of the remarkable aspects of these systems is their ability to interpret and respond to natural language commands, eliminating the need to navigate confusing menus or create complicated formulas. These systems also explain concepts in ways people can easily understand using metaphors and analogies that they can relate to. If an answer is too confusing, you can ask it to rephrase the response or provide more examples.
   
4. Unprecedented scale: Innovations often catch on slowly, as start-ups must penetrate markets dominated by well-established companies. AI stands in stark contrast to this norm. With tech giants like Google, OpenAI and Microsoft leading the charge, the capabilities of large-language models are not only rapidly scaling, but becoming deeply integrated into a myriad of products, services and emerging companies.

These capabilities are finding their way into the classroom through early experiments providing a tantalizing sense of what might be possible.  

• Tutoring assistants: The capability of these systems to understand and generate human-like text allows for . They can offer explanations, guidance and real-time feedback tailored to each learner’s unique needs and interests. and are also piloting GPT-4 powered tutors that have been trained on their unique datasets.
  
• Teaching assistants: Teachers spend hours on , from lesson planning to searching for instructional resources, often at the cost of less time for teaching. As capable reasoning engines, AI can assist teachers by automating many of these tasks 鈥� including quickly , developing worksheets, drafting quizzes and translating content for English learners. 
  
• Student assistants: AI-based feedback systems have the capacity to offer constructive , including feedback aligned to , which helps students elevate the quality of their work and fine-tune their writing skills. It also provides immediate help when students are stuck on a concept or project.

While these technologies are enormously promising, it is also important to recognize that they have limitations. They still struggle with some math calculations and at times offer inaccurate information. Rather than supplanting teachers’ expertise and judgment, they should be utilized as a supportive co-pilot, enhancing the overall educational experience. Many of these limitations are being addressed through integrations with other services, such as for dramatically better math capabilities. Put another way, this is the worst these AI technologies will be. Whatever shortcomings they have now will likely be improved in future releases.

The unprecedented scale and rapid adoption of generative AI mean that these benefits are not distant possibilities, but realities within reach for students and educators worldwide. By harnessing the power of AI, it is possible to create a future where teaching and learning are not only more effective and equitable, but also deeply personalized, with students empowered to reach their full potential and teachers freed to focus on teaching and fostering meaningful connections with their students.

As the director of the Center on Reinventing Public Education, I鈥檝e been closely following the impacts of the COVID-19 pandemic on our education system. One of the most significant impacts has been the widespread school closures, which have disrupted the education of millions of students around the world.

As we start to emerge from the pandemic and schools begin to reopen, it鈥檚 important that we focus on helping students recover from the impacts of these closures. Here are a few key ways that we can do that:

1. Prioritize mental health: The pandemic has been a stressful and unsettling time for everyone, and students are no exception. It鈥檚 important that we prioritize their mental health and well-being as they return to school. This could include providing additional support for students who are struggling, as well as incorporating mindfulness and stress-reduction techniques into the curriculum.

2. Address learning loss: Many students have experienced learning loss as a result of the pandemic, with disadvantaged students disproportionately affected. It鈥檚 important that we provide additional support to help these students catch up, whether through summer programs, tutoring or other interventions.

3. Rebuild relationships: The pandemic has disrupted the social and emotional bonds that are so important for students鈥� learning and development. As schools reopen, it will be important to focus on rebuilding these relationships and creating a sense of community within the classroom.

4. Flexibility: Finally, it鈥檚 important to be flexible and responsive to the needs of students as they return to school. This may mean providing additional support for those who are struggling, or offering alternative learning options for students who may not be ready to return to in-person learning.

Not bad, right? ChatGPT ended that essay with a typical high school writing crutch: 鈥淚n conclusion 鈥︹�� and then a repeat of the same points. You鈥檒l also notice the AI didn鈥檛 recognize that schools are, in fact, open again.

I also asked ChatGPT to write me an op-ed for 社区黑料 on the topic of extraterrestrial education. The essay was coherent and relevant, citing the need for students to develop strong foundations in the science of supporting human life on other planets, environmental sustainability, and 鈥渋ntercultural communication and diplomacy.鈥�

It鈥檚 fun to play with these bots, which can write sophisticated comparative essays and even draft legislation. My communications manager quipped when she saw these essays that she is already out of a job. My colleagues at Arizona State University are wondering how they will assess student essays for plagiarism, even next term. 

If these bots, which are in their earliest and most crude state, can spit out writing like this already, what will they do in two years? How many jobs will disappear in that period alone 鈥� or shift to require radically different skill sets, higher-order thinking and the ability to collaborate with humans and robots? This is a sobering reminder that for CRPE鈥檚 25th anniversary is already here. 

We said the rise of artificial intelligence and other technologies, combined with the realities of climate change and geopolitical dynamics, could lead to rising inequality, civic unrest and other chaos. On the other hand, we argued, the future is still largely within our control. If we embrace change and innovation, we can prepare the next generations of thinkers, problem solvers and leaders to not only confront the complex realities ahead, but to also discover new ways to thrive. We could try to prevent bot-written essays by confining students to pen and paper in a room with no technology. But what if, instead, we asked them to analyze the AI essays to understand why it鈥檚 important to break the rules in writing to communicate more sophisticated thoughts and ideas?

As we kick off 2023, we face the urgency of addressing the pandemic鈥檚 immediate impacts, such as learning loss. But the bot serves as a potent reminder that we will have failed this generation if we do not also use this moment to shift public education toward the future. Here鈥檚 what the bot had to say on that:

The future of work is rapidly changing, and it鈥檚 important that our education system keeps up. The jobs of tomorrow will require a combination of technical skills, critical thinking and collaboration, and our education system must adapt to meet these evolving needs.

This is a daunting challenge, but we and others have been writing and thinking about this for several years. In my , a year before we at CRPE became consumed with pandemic response, I wrote that we needed to more aggressively design schools for kids who are complex learners; radically reshape the high school experience; shift more student supports to out-of-school time and community organizations; make public funds more flexible and longer-term to allow for lifetime education and career retooling; and shift oversight and accountability toward learning pathways and trained customized opportunities, such as tutoring and career training. 

Today, these recommendations seem even more relevant than ever.

A version of this essay originally appeared on the CRPE .