Inside and outside of the classroom, concerns, debates, and tips about generative AI have become the center of our cultural conversation. GenAI is not something to be afraid of; it’s something to embrace and explore. In November, HC at UCLA had the wonderful opportunity of sitting down with Ava Amini to chat about generative AI and how college students can use it to leverage their studies and career paths. As a senior researcher at Microsoft Research, an organizer and lecturer of MIT’s introductory course on on deep learning, and a co-founder and director of Momentum AI, Amini is helping students and patients alike.
As college students, we need to stay engaged with ever-changing technology to maximize the work we do. Many of us are outside of STEM fields, and generative AI is merely a nebulous, slightly ominous concept to us. Luckily, Ava gave us a brief but detailed history of the development of AI and generative AI platforms:
Some Background on AI
AA: So, probably the most mainstream tool right now is something like ChatGPT or Think Chat, and that’s part of a reason why AI is so prominent in our cultural conversation. But the fact is that AI has been around for a very long time, since the 1950s, actually. And in that process of the initialization of the concept of AI to now, there has been a lot of progress; but there have also been a lot of ups and downs in our field, where researchers and our community were really gung ho about particular subsects of AI, and then at times pulled back, and there was a lot of distrust or skepticism. So, it’s not just a linear trajectory straight up; there have been ups and downs.
And for me, I like to think of AI, artificial intelligence, as any sort of computer program or software that is designed to perform tasks that are akin to “intelligent tasks” that we as humans know and design. So maybe that’s something like, in an image, can we identify the objects in that image? This is a task that we now have AI models that are very good at this. And it’s also a task that we as humans do automatically without even thinking twice. And so, that’s kind of my broad definition.
And generative AI is also not a new concept. It’s also been around and developed for many years now. The broad idea of generative AI is now we have AI models or programs that can learn and look at examples of real data from our real world – whether that’s text, images, music, even information about molecules or biology – and find patterns in that information and now generate new instances. And so, rather than looking at an image and saying, okay, this is a cup, this is a plate (that’s a predictive problem), a generative model can actually generate a new image that maybe has chairs, or plates and cups. So, this is a distinction.
The difference is that despite the long history of AI, why it’s so prominent now is that everyone has access to these pretty powerful AI models. We have created this way of interfacing with AI through written language, which is what ChatGPT kind of revolutionized with its public access. So, you don’t need to know how to write computer code, you don’t need to know the intricacies of the math or the computer science behind how AI models are built; you can use written language to work with and use AI models. That coupled with very significant improvement in how good these models are is kind of why it is now such a prominent thing in our world.
College students and generative ai
Now that you have a sense of where AI originated from and why it is currently such a hot-button topic, here is Ava’s advice on how college students can learn to navigate and use these resources.
HC: How do you recommend college students outside of tech go about learning AI when it’s not being taught in the classroom?
AA: For me, with learning anything, I think the two best ways to do that are, 1) to learn the foundations, and 2) to experiment and get hands on, playing with it. With the latter, my advice or guidance would be for students who are not necessarily in tech or STEM to pick something they’re passionate about, an interest of theirs, whether that’s academic or otherwise. Try to figure out and experiment how AI can be used as a tool to guide, enhance, and supplement their creative processes in pursuing that interest. So, what this could look like would be, for example, in the case of designing visual content, you’d use image generation tools to facilitate that, whether that’s Microsoft Designer or DALL-E or something else like that. So, that’s part one, which is just to go for it, experiment, pick something you’re interested in and just play around.
Then, the second thing is for those who are curious and maybe want to learn more about what AI is, how it works, as a way to supplement their experiments and creative process. There’s a ton of great content online that is designed to be accessible to general audiences. And this is everything from blog posts to courses to brief videos, just a lot of different formats of content that people can turn to. As two concrete examples, in addition to research at Microsoft, I’m very involved in this latter aspect of instruction about AI and teaching about AI, and so I have a course called “Introduction to Deep Learning”, which is an online course which teaches the very fundamentals of neural networks, which are the basis of all modern AI systems. And all the content for that is open source; lectures, slides, videos, hands on programming labs if people are more technically inclined; and it’s designed to convey the intuitions and the fundamentals to people from diverse backgrounds and interests.
HC: So, it’s possible to become relatively self-taught in AI?
AA: Yes, I think with everything it depends how deep you want to go. But the thing I’d like to emphasize is that it’s really fast moving technology, and even for someone like myself, who’s in research at a particular niche of the field, it’s moving so quickly that we always have to keep learning about the technology as it grows and evolves. And bottom line above all else, I would say, having that intrinsic curiosity and desire to learn is the most important thing. I do believe that, yes, people absolutely can develop their understanding and intuition about AI because there are so many great resources that we and others have made available.
HC: What is the biggest myth about AI, in your opinion?
AA: When it comes to myths about AI, to me, one of the biggest ones is the myth that AI is meant to mimic how humans think, or to mimic the human mind, which is not true, strictly speaking. Some classes of AI models take some inspiration from how neural systems or the human mind works, but it is not a goal of AI as a field to build an exact simulation of the human mind. And, rather than that, I see AI, especially these types of AI (like ChatGPT) that are prominent in our world, as a tool, as a co-pilot, as a guide that can work alongside humans. It’s not meant to mimic how humans think, it’s not meant to take over all of human capacity, creativity, and work. Rather, it’s something that could function as our guide, even as our guardian, in various types of tasks.
HC: What is your best advice for college students entering the workforce, regarding AI?
AA: I think my best advice is, be curious about the technology; seek to learn more and seek to try and use the technology. To me, with the familiarity and knowledge comes awareness and understanding and power. The technology is going to change, so be committed to continuing your learning as the technology changes. Also, be curious about how you can leverage AI in things that you deeply care about. If you have ambitions in your career or your studies, thinking about, say, different subjects, how can AI help us solve problems in those different areas.
For me in my personal journey, academically, I was very interested in biology and medicine, and then I developed interest in thinking about how we can build AI as a way to answer some of the hardest outstanding questions in biology and medicine. So, think about these cross discipline connections. And the final thing is, be part of the conversation. Be actively engaged because people have power, voices have power. These things are not designed and used in vacuums; they’re a technology that interfaces with society, and all of us make up society.
HC: What do you think the future of work with AI looks like?
AA: AI is a technology, and there are other examples of very strong, computationally driven technologies, like the internet, or smartphones, or laptops and personal computers, that very profoundly changed the fabric of how everyone works and how everyone communicates. And I think that AI can and will be such a technology. And so, it relates back to the theme of learning alongside the technology as it changes, and also being good stewards of the technology in how we work with it and use it responsibly and fairly.
HC: So, it’s not something to be afraid of, it’s something to be curious about.
AA: Exactly.
There you have it. No matter your major, area of interest, or skill set, generative AI is a growing tool that can and should be used. Taking a little time every week to research and experiment can set you apart from those who are resistant to these changes. HC at UCLA was fortunate to have this informative conversation with such an intelligent engineer, scientist, researcher, and teacher. Thank you to Ava Amini for such informative insights on why all college students should be making the most of these up and coming resources. And thanks to you, HC at UCLA reader, for being open to recognizing the importance of developing technological adaptability.