After spending about six very influential decades at MIT Maths Department, in May 2023 Professor Gilbert Strang received a standing ovation for delivering his final lecture to 400+ students. Strang has spent 66 of his 88 years at MIT — as a student, an instructor, and a faculty member.
3+2+61=66, or 75% – summarises the life of Gilbert “Gil” Strang’s at MIT.
Analytics India Magazine got in touch with the maths genius to understand the scope of AI in mathematics and discuss his retirement plans.
“MIT’s computer science department is doing things on a large scale, I just thought the maths department should have a little role. So, I created the newer course that I planned for maybe three or four years and also wrote a textbook for it,” he said talking about the collaborative way departments work at MIT.
In the early 2000s, MIT decided to create OpenCourseWare and instead of trying to sell courses, made them open to everyone. Then they recorded Strang’s ‘linear algebra course maths 18.06’. “In fact, I think India has more people who know that, of course than any other country in the world,” Strang stated.
“Linear algebra has many applications all over and in engineering, but also in computer science, economics, and everywhere,” he said, emphasising the importance of the subject in academia. The mathematics of data undeniably is a subject to study prior to getting started in AI/ML.
“I have to say something more about India,” he continued, “because my whole work depends on my contact in Mumbai [Wellesley Publishers, Mumbai] who typesets all the books, papers, and lectures online. In other words, they do everything that I couldn’t do and is absolutely essential in my teaching.” Strang still writes books by hand and scans them to publishers in Mumbai. His books are published in the US by Wellesley Cambridge Press. Now, Wellesley also exists in Mumbai to sell at Indian prices. “Things are moving in all good directions and it’s just exciting to be including data science along with linear algebra,” he said.
Will AI be as impactful as OCW for Math
The professor, who spent 75% of his life at MIT, first uploaded his classes to MIT OCW, the same year of its launch. “I just thought it would be a good idea to record the lectures. A few months later OpenCourseWare started so it was maybe the first course with video lectures in an open courseware,” he said.
The MIT computer science department is very big and has the largest number of students. Maths is now a second. “I’m glad to say when I was an MIT student, there were about 8 or 10 maths majors; now we’re up in the hundreds. It’s an exciting time so I’ve learned a little bit about deep learning and neural nets,” Strang gladly boasted. He has been in the mathematics department all his life. Many years ago, Strang was handed over the linear algebra course. “Actually, it wasn’t so large at the time because it was very pure. Then, the engineering and computer science students jumped in so that course became large,” he recalled.
Plans Post MIT
“It’s a little different after 61 years of teaching to be not planning for courses in the fall,” said Gil talking about his post retirement plans “but invitations keep coming”.
Talking about the ongoing JuliaCon 2023, where Strang will be delivering a lecture, he pointed out that this year is the 60th birthday of Alan Edelman, his colleague and the creator of Julia. “I’m busy creating my short lecture which will be about the beginning of linear algebra, that’s taken from the first chapters of the linear algebra books. I’m still busy, at least up through next week: preparing for the conference.”
Speaking about the progress towards AGI and the role of maths in it, Strang said, “It’s a natural aspiration. It is inevitable to carry computer science and maths forward. What AI has done already and will do is overall positive. Mathematicians will play not the first part of seeing what can be done, but then understanding what makes the idea work.”
Strang further mentioned that he is an admirer of DeepMind, the AI lab founded by Demis Hassabis and acquired by Google. Excited about their work, he said, “They’re achieving miracles, right? I wonder if a Nobel prize can be given to a whole company for solving the problem of protein folding. Creating the software to be world champion and go and other things are interesting, but that protein folding code is truly a major step in drug development.”
“The key question is, if the data is slightly changed, is the output slightly or largely changed?” Strang pondered.
Answering the question he said, “When the data changes a little the output just changes a little. It’s stable and can be done from a maths point of view. That’s the success of deep learning. Table interpolation. I can’t say it’s fast because the computing requires enormous computing power and costs, but it’s now achieved.”
When asked how AI would affect teaching in the future, he said, “I’ll mention a little bit about something that happened before ChatGPT hit the world. Iddo Drori, a computer scientist I know, published a paper with many authors. He created a software that would read, understand and solve maths problems. He used linear algebra as his first example. That was the beginning of the effect of deep learning on ordinary teaching classes and learning about ordinary subjects like linear algebra,” the professor reminisced.
“I think that’s the first time that computers effectively could understand and create maths homework problems. But now ChatGPT arrives and it overwhelms earlier ideas. We certainly don’t know what’s in the classroom. I don’t feel well-qualified to predict the future there but it’s certainly exciting and important and we’ll see how it develops,” he concluded.