We have translated 4,000 courses in 18 languages, including Hindi, using AI: Jeff Maggioncalda of Coursera

Online learning platform Coursera is going big on artificial intelligence. The company has translated 4000 courses in Hindi using AI, launched Coursera Coach (beta), a GenAI-powered virtual learning assistant, and launched Coursera Course Builder, an AI-powered course-building tool that will auto-generate content. In an interview with Business Today, Coursera CEO Jeff Maggioncalda talks about the company's new AI solutions, how professionals can stay relevant with new-age technologies, popular courses on the platform, and more. Edited experts. 

Since our last meeting in October 2022, when the world was recovering from the aftermath of the Covid-19 pandemic, it has now been dominated by Generative AI. How do you see the ed-tech space evolving with the adoption of the likes of ChatGPT and Bard AI? 

It is changing very, very rapidly. This time last year, the world was just introduced to Generative (Gen) AI. In the last year, even just at Coursera, we have translated 4000 courses into 18 languages, including Hindi, which means that almost everyone in the world can now learn on Coursera. We have created something called Coach, which is a personalised learning assistant, so that if you're having trouble learning the concept on Coursera, you have your own tutor who can teach you and not just teach you about the course but could actually help guide you in your career. We've also built something called Course Builder that allows universities and businesses to use content on Coursera to create custom courses for students and their employees, and we're just getting started. So things are moving very, very rapidly. 

Usually, how long would it have taken you to translate a course in a different language, and how is it different with AI? 

We have translated courses in the past, and it would typically cost $10,000 to translate one course into one language, usually about three months. Now we can translate a course into a language for $20. On Tuesday, I just launched a course that I created called ‘Navigating Generative AI: A CEO Playbook’. I finished my last video on Sunday, and by Tuesday, the entire course was translated into 20 languages. 

But with AI, you have to put lots of checks and balances in place. What checks has Coursera implemented? 

A lot of it has to do with the application of GenAI. What are you using the model for? You raise a great point, though, which is that any user of Gen AI needs to understand not only the general limitations and risks of the technology but also the specific limitations and risks that are relevant to that use case. In language translation, what we needed to do was make sure we put a human in the loop. So before we actually release the course, we have linguists look at the actual writing to make sure that it's up to standard. We also have learner disclosures. Everything is disclosed, including that this is translated by a machine. And we have the ability for learner feedback, which is that if a learner sees something that doesn't look right, they click a little button that says not good. And then we have humans go and fix that. So part of it is the awareness of when there might be an issue by the learners and having a human loop to fix it. 

Now, when it comes to Coach, which is the interactive tutor, there's a different set of risks. In this case, we don't want it to give the wrong answer. The way that we have constrained that risk is that when you're in a course and you ask a question, before we answer the question, we say, Is that question relevant to the course? If it's not relevant to the course, we say that's not relevant to the course, and we're not going to help you with that. Then, when we answer the question, we get the content of the course that came from the instructor. And that's the basis for answering the question. So when people worry about hallucinations, there's something called context grounding, where you can say give the answer from this set of content. So we give the course to the chatbot and say, Make your answer from this expert content. And then, when the answer comes back, we say, again, is that answer relevant to the course? And if it's not, we say we can't give you the answer. So you need to use that technology and understand the risks. 

Where many organisations discourage their employees from using generative AI, Coursera, on the other hand, motivates employees to use the technology. What’s the idea behind this decision? 

There is no doubt about it. Study after study after study,  plus my personal experience as a CEO using gentle AI, software engineers using generative AI, and almost every single job function. It's not theoretical. People using it are realising they're far more productive. But you have to make sure that you use it in the right way. And so teaching people, everybody, the right ways to use it and what the risks are are really important. There's a saying in Formula 1 that says if you want to go fast, you need good brakes. And I think something similar is true. If you want to jump, you'll adopt generative AI quickly. You need to have guardrails, ethical statements, policies, and even technical sandboxes to protect security and make sure that people only use certain language models so that you mitigate those risks. You can't just say, Go use Gen AI. Well, you can, but it's very dangerous. You have to strike a balance between going fast and being safe. 

How is Coursera using Gen AI for its own operations? 

We are doing a lot of things. Our software engineers are using gentle AI for peer-probe pair programming. Our marketers are using the power of AI to create and translate copy. Our salespeople are using generative AI to develop sales scripts and prospecting emails. Our customer service agents are using Gen AI, and this is amazing. We now have agents that speak one language, and they're serving people who speak five different languages. The software takes the learner request, translates it into the language of the service agent, types the answer in their language, and then translates it back in real time with a chatbot. And then for me, I'm using generative AI. So you can use AI to summarise things, help you evaluate things, analyse things, compare things, integrate things, and communicate things, and I'm using it as a thinking tool to help me think more clearly and communicate more clearly. 

When we last met in October 2022, moonlighting was the big issue. Today, because of Gen AI, there is a big problem with layoffs. So how can this be tackled? 

We don't know for sure which jobs will be impacted the most. Certainly, lots of research suggests that jobs that involve spoken language and written language programming languages, as well as audio, video, and images, are probably going to be impacted quite a bit. But McKinsey just came out with a report that says even though STEM professions will have a lot of impact from GenAI, the demand for STEM professionals will still increase. So I think a lot of companies are being conservative right now. They're trying to figure out which jobs I'm going to need and how people will use this differently. I still think, and McKinsey thinks, that there's going to be a long-term demand for STEM professionals. 

How can professionals stay relevant and upskill themselves? 

I think the most important thing to really think about is: in your job, how do you create value? I mean, as a CEO and as an employer, any employee, whether it's a journalist or whether it's someone writing software or a salesperson, is like, How do you add value to this company? And then the question is, how can you create better work output, more effective deliverables that have a bigger impact on customer service or on any part of the company, and then how can you do that more effectively and more efficiently? 

So the first big piece is going to just say, How can I do my job better using these tools? But then the second piece is, how might my job change altogether? In a world where technology might allow me to not have to do these tasks, what are the new tasks that I can do that would be even more valuable? So people often talk about automation versus augmentation. And almost every job is going to have part of it automated. And then the key is for the individual to say, What can I do with my freed-up time that's more valuable for my company? So people who are lifelong learners and really think about delivering value to the customer and value to the company are going to do their best in the workplace. 

If we look beyond Gen AI, there is a huge talent crunch when it comes to sustainability and ESG-related roles. So do you have courses on our platform? And what is the response from India as compared to Southeast Asia or other parts of the world? 

Sustainability is obviously incredibly important because it's at the top of almost every government's agenda. When it comes to investing, investors also care about sustainability and ESG. What we've seen on Coursera, though, is that institutions are very interested in sustainability programmes and certification. We are seeing very good uptake within institutions of how to make their businesses more sustainable by training their employees on sustainability. 

Individuals, though, without being part of an institution, aren't really buying a lot of sustainability courses because it's hard to know, like, What job am I going to get if I learn this thing about sustainability? I suspect what's going to happen, and we're heading in this direction, is that there will be function-specific sustainability certificates, so you won't be the sustainability person. I mean, some people will but you're going to be an operations manager who has sustainability expertise. Or you'll be a production line manager who has sustainability expertise or a software engineer who understands sustainability with respect to computation. Sustainability will be an adjacent skill for almost every job, but individuals don't seem to be going out and buying the certificates as much as institutions are. 

Do you see organisations helping their employees upskill in, say, Gen AI or sustainability-related courses or in some other areas? 

Certainly, with respect to sustainability, that’s a long-term, important agenda item. GenAI is both important and urgent. We just launched the Gen AI Academy in December, and now companies are rolling out generative AI for everyone. That’s an educational programme to teach everybody the basics of the capabilities, the risks, and the ethics. Then we just launched in January gendered AI for leaders, and I have this course, generally AI for CEOs. And then the third piece is going to be gendered AI for team leads, which goes function by function and teaches each function how the generation of AI will change their business. So we see Gen AI going very rapidly. And sustainability is important, but Gen AI seems to be more urgent right now. 

I believe L&T has partnered with Coursera today, so could you throw some light on that? 

L&T is an amazing organization. It is a company that does a very wide range of things. They have Coursera for every single employee, and they set up academies and finance academies in cybersecurity, and now they have the first company in India to implement the Gen AI Academy for all employees at L&T. They also not only use Coursera content, but they also publish content from L&T on Coursera. 85,000 people around the world are learning from L&T because they've published their courses on Coursera. So it goes both ways. 

Which are the top five courses on Coursera globally and in India? 

There's a lot of overlap. I will say that among the top five on Coursera, that's not on India, is ‘The Science of Well-Being from Yale’, very popular internationally. It's popular but not among the top 10 in India. There's another one, which is ‘Learning how to learn’, which is very popular internationally but not as popular in India. The third one is ‘KPop: First Step Kpop’. There are a lot of people in different countries who love K-pop, and they all want to learn Korean, but not as much in India. These are the top courses in India: ' Data-Driven Decision Making’ from Google, ‘Financial Markets’ from Yale, ‘Leadership Fundaments’ from IIM Ahmedabad, ‘Cybersecurity’ from Microsoft, and ‘Machine Learning and Regression’ from Stanford. These are very practical, business-oriented, job-oriented skill programmes, and that is the character of India; people really see education as an opportunity to drive their careers forward. 

Also read: IIT Guwahati Launches Degree In AI In Partnership With Coursera, JEE Not Required