A For AI? Edtech Start-Ups Embrace Generative AI As ChatGPT Sweeps The World

25-year-old Mahesh Prajapati signed up for a Python programming online course with an edtech as he was eager to explore career opportunities in the data science role at his current workplace. However, during the 8-week training program, he often fell back on generative artificial intelligence (gen AI) tools to complete assignments while juggling work deadlines.  

"To my surprise, my assignments were accepted, and I breezed through the programme," he noted. 

Prajapati is among the many students leaning on gen AI and large language model (LLM) tools like OpenAI's ChatGPT to complete their homework and other assignments. While many might call it cheating, the students call it assisted education. 

View it whichever way you prefer, but there is simply no denying that the advent of LLM-based applications like ChatGPT is set to revolutionise how education functions, especially when content and engagement have become the crux for delivering the best results. It has forced a rethink on how students are taught at educational institutions—encompassing physical and online entities.  

A Fitting Lesson

ChatGPT is expected to catalyse conventional education because it could rejig the matrix that placed all learners on the same educational trajectory, which often did not yield the best outcomes. Considering that each learner has their own pace and learning abilities, AI and LLM-based technology enable education institutions to personalise the learning journey for each individual.  

A strong advocate of technological evolution and revolution, Ravi Bhushan, founder and CEO of BrightCHAMPS, believes that gen AI-enabled tutors that support learning through Socratic dialogue and debate will radically level the playing field for students from different social and economic strata. “When AI and extended reality (XR)—both powerful technologies individually—are combined, it will lead to the possibility of learning experiences that can be adapted to individual learning needs, abilities, styles, and interests, making learning a more joyful, long-lasting, and effective process,” he claims. 

Taking advantage of the newly available technology, upGrad developed an interview bot that puts custom interview questions based on learners' performance on non-graded assignments, allowing for a tailored experience that meets their specific needs. Similarly, it is working on creating a coaching bot as well.  

"The education system can benefit from the presence of AI and LLMs for several reasons, as these technologies can provide a personal tutor-like experience for each learner,” explains Mayank Kumar, co-founder and managing director of upGrad. "To effectively incorporate AI and LLMs into the education system, it is crucial to harness these technologies for learning and developing applicable products. Nonetheless, significant efforts are still needed to make it production-ready and ensure its practical implementation across levels." 

In May this year, Amity University Online introduced Professor AMI, which it claimed was India's first AI professor in online higher education. Ajit Chauhan, chairman of Amity University Online, said, "When it comes to online learning, personalisation is the key. Professor AMI will facilitate a focused, on-demand, and real-time learning experience." 

Powered by Open AI-driven ChatGPT technology, this system can enable a personalised learning experience for the institution's students based on their strengths, areas of improvement, and learning styles. In addition to delivering recorded video lectures and responding to academic and learning management system-related queries, it also offers 24/7 chat assistance to students, another area where it scores over traditional educational styles.  

In June 2023, Byju's jumped onto this bandwagon by introducing a new suite of three generative AI models—BADRI, MathGPT and TeacherGPT—under its WIZ suite of learning modules. After understanding students' learning patterns, it can develop a methodology to help them learn better, the edtech’s co-founder Divya Gokulnath told PTI. 

Harbinger Of Doom? 

Does the rollout of these tech tools mean that educators should be worried about their job prospects? Many industry experts say it is too early to make these doomsday predictions. However, they are emphatic that these technologies, including AI-powered systems, are designed to complement and enhance the learning experience rather than replace human educators.  

Krishna Kumar, founder and CEO of Simplilearn, notes that ChatGPT's capacity to understand context and deliver human-like responses is one of its key features, making it well-suited for usage in chatbots, virtual assistants, and other conversational applications. Understanding the technology and its applications is important for professionals across various sectors, not education alone.  

"Gen AI can accelerate and support functions in ways that may not have been imagined previously. However, it will not threaten the livelihoods of education professionals as long as they remain agile and can adapt to new ways of work. This requires them to invest the time to learn and understand the new technology to use it effectively in their professions," Mayank claimed. 

While these tools can provide personalised tutoring and support, the crucial need for human guidance, mentorship, and the cultivation of critical thinking skills can't be displaced. Simply put, the role of educators will continue to be vital in creating meaningful learning experiences and fostering intellectual growth in learners.  

Hence, rather than viewing these technologies as harbingers of doom, educators should consider them as allies. As virtual teaching assistants it can aid teachers in lesson planning, addressing individual student doubts, providing data-driven insights for customised instruction and automating administrative tasks. This, in turn, will offer them enough time to focus on more impactful work and foster analytical thinking skills among teachers. 

Some edtech companies like Udemy have already embraced LLMs, layering them with ChatGPT to help language teachers design their courses. upGrad introduced a multi-layered GPT integration for its learners. They can select curriculums and live classes to understand the tool's architecture and fundamentals and utilise the capabilities to drive innovation. It also helps in assignment grading.  

It is unsurprising to see edtech start-ups pay closer attention to ChatGPT and other LLMs. At a time when they are battling a funding crunch, competition from offline schooling and declining revenues, these technologies can help them prune their operational cost and harness the potential of their greatest asset—their faculty.  

Edtech start-ups can explore opportunities to create innovative products and services incorporating LLMs, such as AI-powered tutoring platforms, virtual learning environments, or adaptive learning systems.  

A Note Of Caution  

While adopting this technology appears to be a win-win, the real test lies in how well the faculty is trained in using it while maintaining academic integrity. Moreover, given the growing noise about the importance of responsible AI use, they must prioritise compliance with regulatory guidelines while addressing concerns like bias, privacy, and security. 

Since LLMs extensively bank on limited databases, they tend to hallucinate, which means spewing incorrect information. The accuracy and reliability of responses it provides to learner questions depend on various factors, including the available database and the model's ability to generalise information.  

Fortunately, the newer LLM-based models are putting guardrails to combat these hallucinations and provide access to external data. Edtechs can tackle LLM hallucination by choosing its optimum degree and tweaking this in different curricula.  

For courses that require creativity and inspiration, such as coding, designing, or other creative thinking disciplines, they can select a higher degree of hallucination while training the model with custom data. This allows for more innovative and imaginative responses while still maintaining educational integrity. 

LLMs should also be used in cases where the accuracy threshold is above acceptable levels. For instance, it can be valuable in fields like coding, where precision is crucial since it can provide accurate and reliable guidance, offer solutions and assist in problem-solving. In such cases, LLMs can enhance the learning experience by quickly and efficiently addressing coding-related queries. 

Krishna also cautions that LLMs should serve as educational tools rather than replacements for the entire education process, and this approach requires careful consideration and design. "Another important facet is to help learners understand the limitations and biases of LLMs and provide them with the necessary context to critically interpret and evaluate the responses generated by the model," he adds. 

Bhushan notes that the biggest innovations and leaps that humankind has made have come from times of friction and churning, not from stagnation or toeing the line. “Does the integration of AI within learning and education bring up some sticky subjects? Yes. Does it mean we live in fear of tech? Absolutely not,” he emphasises. 

The writing on the wall is clear: cracking down on generative AI and LLMs tools will not help educators, students or the edtech sector at large. And when you can't beat it, it is better to join it. 

Rather than considering the use of AI to manage assignments as cheating, it's better to teach them how to use it optimally, given the wealth of additional resources and knowledge it offers. Because, love it or hate it, ChatGPT is here to stay. While it may change how students study and teachers teach, it will not be able to replace the relationships and peer-to-peer interactions that are critical to education.