Jeff Maggioncalda, Coursera CEO: A Fortt Knox Update

Well, it's a few things. I think that for companies, it has taken longer than I would have thought for them to get their arms and their heads around this whole thing called generative AI. What people are realizing is that learning about AI has always been valuable in the past, but mostly just for the people who build AI. With generative AI, there's a whole new opportunity for companies to teach people who use AI as a tool, as a thinking tool, to become more productive. And so companies are now really shifting from saying, 'What's our gen AI strategy?' to 'How do we actually unlock productivity by teaching people in different job roles how to use gen AI to do their jobs better and more efficiently?'

I think it's a great question, and I haven't really thought about that. What are other analogies? I often talk about gen AI as a thinking tool. It's not just a search engine. I mean, clearly with Google Gemini, they've done a nice job making it so that you can ask the AI questions, but then it does a web search and it summarizes the answers, so it's becoming quite good as a search engine. But the real power of gen AI is the ability to transform thought, to combine ideas. So you could combine your own perspective with one that you read about, you could combine company strategy with an initiative that you're working on. But this ability to transform ideas and transform thought is what's really valuable. So there are general purpose skills of how do I approach my job differently because of this technology. I think it's not like a calculator. A calculator is a very clear purpose-built tool that says, instead of doing math on a piece of paper, I'm just going to do it on a calculator. But some people call gen AI a GPT, a general purpose technology, something like electricity or maybe something like a computer or something like the internet. At first, people just don't know exactly what to do with it, and then suddenly there are certain use cases. So like you said with a WYSIWYG, what's a mouse? 'Oh, look at this thing, I'm moving my cursor.' But then who thought that a mouse would become an input device for drawing on tablets or for controlling Zoom meetings? These new technologies that will be used in ways that we almost can't imagine require people to first get familiar with the basic technology and then move towards how do I get the most value by applying it in certain use cases. And we're just starting that application process with generative AI. And that's shown in the fact that prompt engineering has become a popular thing that people want to learn.

A lot of it is self-reported from the learners. A lot of it is we see trends on the consumer side. You know, what do individuals come to learn on Coursera? General prompt engineering, prompt engineering with ChatGPT, those things are pretty popular. But on the institutional side, we're starting to see more focus on job-specific roles. Like companies are starting to say, 'Hey Coursera, we want you to train our software engineers on how to use, whether that's a co-pilot or some other kind of AI coding assistant, in order to do this.' And what we're also doing at Coursera is making it so that as you do your programming exercises, AI, our little coach assistant, is kind of helping you and giving you feedback on your coding as you write code. So coach can be very job and task specific and giving you feedback. And then the content that we're bringing on from our partners is also job and task related. And generally speaking, what we see in terms of the jobs that are really being impacted the most earliest, it's customer service jobs that use a lot of language. They are learning how to use this sort of augmented capabilities rapidly. There's a lot of productivity gains there. Marketers are using it like crazy. Salespeople are totally changing the way that they pitch to prospects. Software engineers are coding more efficiently. And data analysts are doing a lot different. We also see a lot with people skills. A lot of HR teams are starting to use this as well. There's a lot of words, there's a lot of language associated with the job tasks in HR. So role-specific gen AI training is becoming a big thing.

Yeah, we just did this job skill report. And what this is, it's an analysis of all the learners who use Coursera as part of an institution, so employees whose businesses have hired Coursera, or students whose universities have hired Coursera, or governments, citizens whose governments have hired Coursera. So it tilts a little bit more towards what do institutions want people to be learning. And the list is the fastest movers in terms of skills, which skills are gaining in popularity. And clearly there's a lot of gen AI, but to your point, there's data analysis, there's cybersecurity, there's business optimization, there's risk management. And I think the overall theme is adapting to change. Right? There's a lot on cybersecurity. Well, when you have distributed workforces working remotely in many different countries with a lot more adversarial cybersecurity threats coming at you, training people on how to manage networks and protect those networks against cybersecurity threats turns out to be a big thing. On assertiveness and the soft skills, a lot of what that is about is how do you manage and lead people through change, how do you adapt to change as a team member. So communication skills, there's a lot on project management, a lot of interest in project management, a lot on risk management. A lot of what happens is when things are changing quickly, it's not clear where the risks are, and so we're seeing an increase in risk management. So I'd say generally speaking, the package of skills is kind of those skills most associated with coping with change are the ones that seem to be growing the fastest.

Yeah, I think that generally speaking, the companies are going on an adoption curve that looks something like this. Point number one, or phase one, is awareness: 'Oh my gosh, generative AI is a big thing, I should pay attention.' Point number two is generally trying to figure out strategy, which often includes risk management. So they're like, 'Uh oh, what do we do as a company? How might this change our business model? Which job roles and tasks and processes might need to be re-engineered?' Etc. And then the third phase is kind of implementation. And I think that after talking a lot about ChatGPT and then hiring a lot of consulting firms in 2024 to do strategy studies, and still in certain countries waiting for the regulators to say what you can and can't do, I think 2025 is going to be a lot about implementation. And I'm looking forward to a lot of skill training happening as part of that. We're also seeing this coincide with a major move, not coincidental, towards what people call skills-based organizations. These are companies who are saying, 'You know what, I really need to make sure is not that you're just learning in general, but you're continuing to learn the new skills that are required in a changing world. And I'm going to, for your job role, I'm going to help you understand what those skills are that you need to learn, and then I'm going to give you a badge or some kind of a certification so that I know that you've got those skills.' So I think that skilling is becoming more mission critical, and recognizing those skills through credentials is becoming a more common practice as the world changes faster and faster. And I think 2025 is going to be about implementing skilling to adapt to change, starting with generative AI.

I think it definitely should take into account. And for us at Coursera, and what we see a lot of the forward-thinking companies doing, is starting with responsible AI principles. Like, 'Here's how we're going to use AI responsibly.' Generally speaking, as a rule of thumb, what I sort of see is if there are high-stakes decisions that can have potentially unforeseen impact on different populations, it's okay to use AI, but you've got to have a human in the loop who is responsible for the decisions and the impact, and you've got to monitor that impact. You can't just assume that everything's going to be okay. To your point on HR, I think if there is backlash, it would be appropriate if companies were adopting AI for high-stakes decisions about who they hire, and even upstream from that, who they screen, who they promote, what's written in reviews. If there's not a human responsible not only for the decision but understanding the use and impact of the technology in the decision-making process, then I think there is something that people need to address. And that's why I do think that training people on ethics, and in a course that I put together for leaders I talk about ethics and gen AI as a form of risk management, like where the risk is the risk to society, you got to make sure that you're anticipating where the risks might manifest and how you're going to detect them and mitigate them. And I think it's important, especially in HR, because those decisions impact people and their livelihoods. It's important to make sure that people are clear and adopting these responsible AI principles.

Yeah, absolutely. You know, one of the things that we're seeing in Coursera for Campus, we have a number of features for what we call verified learning, where making sure that people aren't cheating, or at least trying to detect and deter people from cheating. One of the things that we do, we just launched this this summer, is that when someone puts in a written submission, this is in Coursera for Campus where a student is putting in like an essay, rather than using AI to try to detect whether the student wrote the essay, our little coach pops up and interviews the student about the essay they wrote. And I think that we might have those kinds of things in the recruitment pipeline where AI will help filter these things out. But it really is important, I think, for the humans who own that process to make sure that the composition, the demographics, the kinds of candidates that are making it to different stages in the screening process, that there's no implicit bias that's showing up because of the way this technology is being used.

You too, John. Take care as always.