These days, the argument regarding the deployment of tools like ChatGPT is likely reaching you when you hear about students and generative AI. Do they provide assistance? (Happy! fantastic for study! Quickly!) Or do they cause harm? (Unfortunately! False information Belittling!). However, some businesses see the introduction of generative AI into the classroom as a good thing that is inevitable. Additionally, they are developing items in response to what they anticipate to be a specific market opportunity.
One of them has now raised some funds to fulfill that goal.
Leading a $15 million Series A investment, MagicSchool AI is developing generative AI solutions for educational settings. Bain Capital Ventures led the deal. The CEO and creator of Denver-based MagicSchool, Adeel Khan, stated in an interview that the company started off making tools for instructors and that today, about 4,000 schools and teachers use its products to create lesson plans, create assessments, and create other learning resources.
It has also begun to provide tools for kids, which are supplied by their schools, in more recent times. The money will be used by MagicSchool to carry on adding to both of those projects, as well as to work on expanding its clientele, employing staff, and other endeavors.
Important investors have also backed this most recent round. These include Common Sense Media, the expert in age-based tech reviews, and Adobe Ventures, whose parent company Adobe has been utilizing AI extensively on its platform. Common Sense Media has also been experimenting with generative AI, partnering with OpenAI on AI guidelines and rating chatbots. Those involved in the round include Rafael Garcia and Tyler Bosmeny, co-founders of Clever, Amjad Masad, creator of Replit, and Amir Nathoo, co-founder of OutSchool. (A portion of these were also the company’s initial seed investors; it had previously raised about $2.4 million.)
The investors feel that supporting application investments like this one is the logical next step for AI businesses after hundreds of millions have been invested in infrastructure companies like OpenAI, Anthropic, and Mistral. Khan did not disclose MagicSchool’s valuation in this round.
Partner at Bain Capital Ventures Christina Melas-Kyriazi stated in an interview that “there is an AI moment for education, a big opportunity to build an assistant for both teachers and students.” “They can assist teachers with lesson planning and other tasks that require them to be away from their students here.”
AI Preacher After Becoming A Teacher
Despite its name, MagicSchool was not created overnight.
When Khan first graduated from college, his first job was as an educator for Teach for America. (And it’s possible that his interest in public service and the importance of education predates that; at Virginia Tech, he was president of the student body during the shooting, so he tragically witnessed the devastation caused by gun violence up close.)
In his early career as a teacher, he shown a knack for leadership and entrepreneurship when he relocated to Denver to pursue his dream of opening his own school.
After taking on various administrative positions at nearby schools, he eventually established DSST: Conservatory Green High School, a charter high school, whose first class of graduates was accepted into four-year institutions with 100% acceptance rate.
It was during that flurry of activity that Khan had his eureka moment for MagicSchool.
He remembered that “generative AI came into the ether for most of the country around November of 2022, when ChatGPT was dominating the headlines.” “I began experimenting with it as I was considering my next move, and it dawned on me right away how useful this new technology could be for educators.”
He visited the schools where he had previously taught himself and showed his former colleagues the potential of employing generative AI to create tools for teachers. But nothing was connecting.
He claimed that “it was just not sticky and the interface was clunky for them.” The intended “wow” was sparked by Khan’s demonstrations for them, but if left to their own devices, the teachers would only use it once.
They would say to me, “It ended up costing me time instead of saving me time, because I spent so much time trying to prompt it and get it to do what I wanted.”
He came up with more targeted adaptations as a solution.
“Behind the scenes, we were just making sure that the outputs were what an educator would expect and doing some really sophisticated prompting,” he explained.
Lesson plans, assessments, quizzes, course materials, and recasts of prepared content for increasingly and less difficult learning levels are a few examples of what instructors are producing using MagicSchool. MagicSchool is still fiddling with this. According to Khan, it integrates well not only with Anthropic but also with OpenAI’s APIs. He explained that the business runs AB tests in the background to figure out what functions best in various situations.
However, it wasn’t exactly easy to get schools, who pay for the service, and teachers, who weren’t paying to utilize MagicSchool, to join up.
“There was so much fear about it all when we started the product that I couldn’t get a meeting with any school or district, including the one where I worked,” he remarked. To put an end to any discussion, all it required was “a negative headline about the use of AI in schools… about how AI is going to take over the world and robots.”
That began to progressively shift as more sophisticated models were released and AI was widely accepted by industry and society. He noted that the primary benefit of using it was time savings, but other benefits included idea generation and even serving as a complement to self-taught knowledge.
He asserted, “I think educators were underestimating the benefits AI could provide to both them and the audience.”
Furthermore, he provides a second case for why increasing AI in the classroom makes sense: since AI will permeate every aspect of daily life, it is the responsibility of educational institutions to ensure that their students are prepared for it.
Though Intelligent, AI Is Not “Human Smart”
However, there are restrictions on the use of AI in all contexts, including educational settings.
“Human intelligence and artificial intelligence are not the same thing. Emergent intelligence in humans is somehow the result of millions of years of natural selection’s pruning. It has a strong holistic component. According to Mutlu Cukurova, a professor of education and AI at University College, London, where a multiyear research lab explores the many combinations of AI and learning, “it is very flexible, cognitively.” (One really practical finding from a recent paper: a hybrid strategy combining AI and humans is required.)
Instead of emergent intelligence, AI has designed intelligence. It is therefore intended to achieve a very precise aim, or group of goals. Though it’s a different kind of intelligence, AIs are extremely good at this specific task and show considerable indicators of intelligence.
This may be especially pertinent to students and how they will learn in an AI environment, or to educators who may lack the expertise to recognize when an AI-generated learning resource, such as an exam, is inadequate.
Automation of some jobs can be a useful use case, according to Cukurova, but “where it becomes problematic is when teachers…do not have enough experience before learning how to do these kinds of things on their own.”
MagicSchool, according to Khan, tries to keep this in mind, especially with relation to pupils. According to him, schools decide which resources to make available to pupils on the platform, and it’s evident when they’ve utilized MagicSchool for a homework assignment.
In principle, this all sounds excellent, but in practice, stress tests may be the only way to find the flaws.