AI Strategy and Startups with Professor Mike Stonebraker

Written By: Audrey Woods

In our modern era of revolutionary technological advancement, Artificial Intelligence (AI) and Machine Learning (ML) appear poised to reshape many fundamental aspects of our society, including how we live, work, and interact with each other. Already, these technologies have empowered businesses to make more informed decisions, optimize processes, and create personalized experiences that resonate better with consumers.
However, not all businesses are ready for the change. Many traditional companies rely on legacy frameworks and outdated thinking that puts them in danger of falling behind. While it might be daunting to keep up with this dynamically shifting technological landscape, the consequences of not integrating new solutions or adapting to the changing needs of consumers presents long-term economic risk.
With his experience as an educator and his involvement in founding ten software startups, MIT Adjunct Professor Mike Stonebraker offers unique insight into the importance of having a ML strategy, the common pitfalls companies are running into, and what he sees on the AI/ML horizon.  
A Specialist in Database Systems  
Professor Stonebraker’s time in academia began with a bachelor’s degree in electrical engineering from Princeton in 1965. He went on to earn an engineering master’s and then a PhD in computer information and control engineering from the University of Michigan in 1967 and 1971, respectively. Immediately upon graduation, he joined the faculty of Berkely as an assistant professor and began his pioneering research on relational databases.
While at Berkeley, Professor Stonebraker invented INGRES (Interactive Graphics and Retrieval System) and Postgres (Post-INGRES), which featured early implementations of the relational model in databases, transforming an academic theory to the default choice in data processing applications. This work, in addition to his entrepreneurial efforts, later earned Professor Stonebraker the 2014 Turing Award, also known as the “Nobel Prize of Computing.” He joined MIT in 2001 as an Adjunct Professor where he has continued to study data management, visualization, storage, and processing. He now helps lead the Data Systems Group at CSAIL along with Professor Sam Madden and Associate Professor Tim Kraska.
Altogether, Professor Stonebraker has founded ten startups so far over the course of his career, which include: Ingres Corporation, Illustra, Paradigm4, StreamBase Systems, Vertica, VoltDB, Tamr, and his latest venture, Hopara, which offers agile digital twins for real-time system visualization.
Hopara: A New Way to Visualize Data
As Professor Stonebraker tells it, he’s always been a “huge fan of Google Maps and Google Earth” because of its on-demand detail and the intuitive way users can move and click around to find what they’re looking for. “There’s no user manual for Google Maps,” he says in conversation with CSAIL Alliances. “It’s obvious how to do it. The only problem with Google Maps is that it only works on geographic maps.”
This led him and fellow researchers to think about a user-friendly system which could visualize the inside of buildings or locations such as factory floors or lab spaces. This research—originally called Kyrix—became the foundation for Hopara, which gives companies the tools to create either 2-D or 3-D representations of their assets, allowing them to monitor locations, devices, and machinery in real time and navigate their individualized “canvas” in the same way they might use Google Maps.
On Hopara’s website, Professor Stonebraker says, “tomorrow’s competitive advantage will go to those who most quickly (and easily) convert data-collection into business value.” This advantage will be facilitated by innovative solutions like Hopara’s platform or by utilizing other AI/ML techniques to prepare for the coming change.
AI & ML: The Definition of Disruptive  
In 2019, Professor Stonebraker gave a talk at the Open Data Science Conference in Boston talking about the biggest blunders that companies are making when it comes to big data. Near the top of the list was the mistake of not planning for AI and ML to be disruptive. What does he mean by that? He says, “for example, software engineers have pretty much all decided to use chatGPT, Copilot, or one of those tools, and the ones I’ve talked to report a 30% productivity improvement.” Put more directly, “if you're in software development and you're not using ML techniques, you're not going to survive.”
These technologies, he explains, will affect more than just the computer science world. “I can't think of anybody that isn't going to be impacted,” Professor Stonebraker says, which means that everyone in the workforce—from employee to CEO—should be thinking about ways to incorporate ML. Of course, programs like chatGPT are one way to do that, but Professor Stonebraker recommends thinking more broadly about the issue, saying, “my point of view is that chatGPT is good at certain things and not very good at other things, and there are other ML techniques that are also very good at other things. ML is a basket of stuff which collectively will be extremely disruptive in my opinion, and chatGPT is just one piece of it that's gotten a lot of press recently.”
The Problem of Silos
Another important issue that Professor Stonebraker sees, especially in large businesses, is that of “siloing,” where communication breakdowns between systems, departments, and even individual employees keep data isolated and hinders company progress and cooperation. While he admits that it is very hard to overcome this issue politically and logistically, he says, “if I was a CEO, I would be trying very hard to stamp out silos, and that doesn't seem to be happening.”
The solution, he imagines, will be disruptive ML technologies that are “so obvious you have to do it,” giving companies the tools to share information more seamlessly and provide a better customer experience. A large part of the siloing issue boils down to data problems—language barriers, transfer issues, storage concerns, visualization challenges—where Professor Stonebraker’s work is poised and ready to make a difference.
Going Forward
When looking at the future of the computer science field and commercial market, Professor Stonebraker has several concerns that he sees playing out in real time. For one, he’s troubled by what he calls the “arms race” of talent acquisition, pointing out that innovative startups and even traditional companies not in the software development space are struggling to lure in the best minds. Another thing he’s worried about is income inequality, which technology like AI/ML has the potential to exacerbate. And looming largest in his thoughts is the threat of climate change, which he says we aren’t doing nearly enough to solve.
However, he has hope that, at least with problems such as climate change, “technology may be able to help.” And where better to be a part of the solution than at MIT CSAIL? He describes the Stata Center as a “cauldron” for innovation, elaborating, “it's a fabulous place to hang out and come up with ideas.” Talking to students, working with companies, and interacting with real-world users helps Professor Stonebraker fuel his research, inspire future companies, and fortify an ongoing legacy of innovation.
Learn more about Professor Stonebraker’s research on his CSAIL page.