Insight from someone who knows what they are talking about.
INTERVIEW | by Victor Rivero
Dr. Satya Nitta is the CEO and co-founder of Merlyn Mind, a deep tech generative AI company working on AI assistants and platforms. A technologist and a business leader, he was previously the global head of AI solutions for learning at IBM Research and led teams in the development of conversational systems, speech recognition, and natural language understanding. Dr. Nitta received his PhD in Chemical Engineering and holds over 100 US patents. He was named as the IEEE Ace “Innovator of the Year” and has received several technology awards for his work. He is a board member of the Biological and Chemical Engineering Department at Rensselaer Polytechnic Institute. He has published over 40 publications and has over 5,000 citations of his technical work. I first met him at FETC at the beginning of this year and what was the beginning of an AI frenzy in education. As the AI story unfolded, we kept in touch and caught up more recently to bring you his insight on technology for learning along with some much-needed perspective.
You’ve been in AI long before the latest craze. Tell our readers about your background and how it has informed your current approach to using AI for educational applications.
We’ve been on this AI journey at Merlyn Mind for nearly 6 years now, long before AI was the zeitgeist. Our mission has always been to build AI platforms and assistants to change the way the world works with technology. Our first solution is Merlyn, an AI voice assistant focused on improving the teaching and learning process in classrooms.
‘Our mission has always been to build AI platforms and assistants to change the way the world works with technology.’
Going back to the early 2010s, my fellow Merlyn Mind cofounders and I were leading IBM’s worldwide team on AI solutions for learning. At IBM, the scope of work ranged from developing machine learning and natural language processing solutions to initiating R&D projects in virtual reality, augmented reality, and cognitive neuroscience.
When we started Merlyn Mind in early 2018, very few people were talking about practical AI solutions. Two big influences totally flipped the narrative: computing power and large language models.
The field of computing itself has advanced very rapidly, and that’s been the engine. Bigger and bigger machines. Faster and faster processors. Faster GPUs. That’s allowed us to process massive amounts of data very quickly. The AI algorithms that we’re using today are not necessarily new, but they’re succeeding because of the massive underlying computing power that we have.
Large language models are essentially large-scale pattern-recognition machines. They can’t really think. They can’t really reason. But we feed them with enormous amounts of data, and with some brilliant engineering – and that’s the real intelligence at work here – they can predict, summarize, and generate text from that data.
Help our readers to better understand the potential of AI, and perhaps allude to or hint at a December launch you are working on – or tell us a few details that you can – about this.
We envision that we will live in a world where every learner, educator, and professional will have their own AI assistants that will be capable of remembering their preferences and learning patterns as well as performing related actions within the tools and resources they use. Our Merlyn AI assistant is a step toward realizing this vision.
To build useful and generally intelligent systems, we are working at the leading edge of AI and with the AI research community to develop models and platforms that will plan and reason, have long-term memory, have better models of human cognition, have a better understanding of the world, and will perform actions on behalf of their users. We believe that over time, generative AI will transform the world as significantly as the birth of the internet itself has.
‘We believe that over time, generative AI will transform the world as significantly as the birth of the internet itself has.’
However, like other AI advances, the most meaningful solutions in a vertical domain like education will come when teams develop AI in a purpose-built way. These platforms and solutions will be imbued with a deep awareness of domain-specific workflows and needs and will understand specific contexts as well as domain-specific data. When these conditions are met, generative AI will utterly transform industries and segments, ushering in untold gains in productivity and enabling humans to reach their highest potential.
Merlyn Mind is one of the only deep tech companies developing platforms and assistants for education, and that’s a key part of who we are. We have the ability to innovate across the entire compute stack.
In the fall, we’ll be releasing the next generation of Merlyn, our AI assistant for the classroom. Like our first product, this next-generation Merlyn allows teachers to operate their front-of-class display, laptop, and internet applications from anywhere in the room. On top of that, the new release will see the introduction of trustworthy generative AI for education. Built for the unique workflows and safety needs of education, our generative AI platform enables teachers and students to have a generative AI experience that retrieves content from our education-specific large language models, not from the entirety of the internet. The result is an engagement that prioritizes safety, curriculum-alignment, and low hallucinations.
Finally, the form factor is greatly reduced. Our next generation of Merlyn is software-based, allowing any educator with a computer to leverage the power of the Merlyn AI assistant.
We’re really excited about the possibilities Merlyn can unlock. With Merlyn, the tech disappears into the background so teachers can do what teachers do best – maintain the flow of class, increase engagement, and drive successful learning outcomes.
Merlyn Mind has already worked with other edtech partners – what role and what place do you see Merlyn having in the greater edtech ecosystem?
In late June, we open-sourced three large language models purpose-built for education. We hope that the developer community will use our models to check the safety of their LLM responses as part of their solutions.
Merlyn is also available in a familiar chatbot interface that responds multi-modally. We are also being asked to make Merlyn available through an API and are in talks with several potential edtech partners who want to build safe and responsible AI experiences into their products using our generative AI platform and purpose-built LLMs for education.
What trends are you looking at within AI in education, and why those?
At the front of our minds, research, and product development is pedagogical support. Beyond AI platforms and assistants, we’re working with world-renowned cognitive scientists to instruction-tune language models based on specific principles of pedagogy that are designed to boost teaching and learning. Drawing extensively upon peer-reviewed research, these principles and strategies emerge as suggestions during the use of Merlyn in real time, and when activated they can further engage learners and unlock plasticity in the human brain.
‘…we’re working with world-renowned cognitive scientists to instruction-tune language models based on specific principles of pedagogy that are designed to boost teaching and learning.’
What message would you like to impart to the overall edtech industry about the future of learning?
AI is going to forever change teaching and learning. But there is so much noise in AI right now and no shortage of businesses popping up that are little more than an API skin over ChatGPT. Are these solutions right for education? Do they collect personally identifying information? Is a school’s data compromised? Is the content retrieved safe and appropriate for the classroom? Do these solutions truly improve teaching and learning? Getting AI for education right is up to the choices districts make today and the tools they allow into their classrooms, and these are the kinds of questions education leaders should be asking.
Merlyn Mind is different. The work that we are doing is foundational, and our platforms and assistants are purpose-built for specific domains.
‘And although we’re still at the beginning of this journey, it’s critical to expose today’s students to this transformational technology now.’
And although we’re still at the beginning of this journey, it’s critical to expose today’s students to this transformational technology now. In a few years, they’re going to graduate and be expected to interact with and leverage the power of AI as they work, serve, and compete in the global economy. Students who don’t have access to AI and learn how to use it will get left behind.
Victor Rivero is the Editor-in-Chief of EdTech Digest. Write to: firstname.lastname@example.org