Ramin Hasani, an MIT AI Scientist, pioneers the field with his liquid neural networks, aligning AI with human values and advancing adaptability post-training. His innovations impact multiple sectors and offer valuable insights into the future of AI.

Profile

Name: Ramin Hasani
Age: 34
Title: AI Scientist 
Company / Project name: MIT 
Area of Expertise: Artificial Intelligence
City of Residence: New York / Boston / San Francisco
Vision/Philosophy/Motto: effective accelerationist (e/acc)
I am BOLD because... I build artificial general intelligence.
First thing that comes to mind when you think of Austria? My second home.

How to be innovative 

Tell us about your professional background. Are there any milestones you are particularly proud of?

I’m an AI scientist at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT. I have been with MIT the last 6 years of my life. I design powerful AI systems that are highly aligned to our human values. I invented liquid neural networks, a class of brain-inspired machine learning systems that can stay adaptable even after training. Today, liquid AI systems are ubiquitously being used for modeling financial markets, predicting diseases, performing natural language processing, and enabling autonomy.

What has been the most valuable lesson life has taught you thus far?

Persistence and determination are key to success.

How does being a part of the BOLD Community influence you and your work?

It means a growth of network, communication, and discussion of new ideas.

What strategies do you implement to cultivate a culture of innovation within your organization?

I encourage people to always continue learning and to think outside the box. I focus on diversity and inclusion and try to find a balance between depth and breadth.

Is there a particular belief or philosophy that guides your approach to life and work?

Effective accelerationism.

What do you think people should learn from each other to foster innovation and collaboration?

I think what people should learn from each other is knowledge and expertise, communication and collaboration skills, creativity and problem-solving skills, risk analysis, and flexibility. These aspects are crucial for fostering collaboration.

What qualities do you believe will be crucial for success in the future of work?

- effective accelerationism
- data literacy
- analytical thinking
- continual learning

Can you share an instance where a failure or setback led to a positive outcome or growth?
A fundamental tool for achieving positive outcomes is the art of storytelling. I remember back in 2016-2017, when we first invented liquid neural networks at TU Wien, it was so hard to get the idea to stick within the AI community. Over the years, we learned that the main problem was nothing technical but was the sub-optimal way we were positioning the idea. Ultimately, with continuous iteration, we told the right story which led to many impactful publications in Nature and Science magazine families, highlighting the tremendous impact of our tech for the safe deployment of AI systems in our society.

What is a BOLD decision you made in your career and why?

I just quit my main job, as a principal AI scientist, at the second most powerful financial institution in the world, the Vanguard Group (with $8.2 trillion assets under management), to start my own venture on building the AGI we deserve! Embarking on this new entrepreneurial journey to advance the science of AI and develop AI for the better excites me the most.

Explain your innovation

Can you explain your innovation and why it matters?

The deployment of modern AI systems poses important socio-technical challenges, such as safety and accountability, in our society. My technology, liquid neural networks, alleviates a large portion of these challenges. Watch a TEDx talk of mine here: https://youtu.be/RI35E5ewBuI

How did you get involved in this? 

My PhD advisor Prof. Radu Grosu from TU Wien introduced me and my colleague Dr. Mathias Lechner to the fascinating nervous system of the roundworm, called C. elegans, which, with a tiny nervous system, can achieve massive degrees of complexity. We studied together the principles of computations in the worm’s brain,and developed liquid networks as a robust and more powerful AI system to model data of sequential nature. The rest is history.

How can others contribute to this effort?

The core technology behind liquid networks is now an established class of machine learning systems which is openly available to AI and machine learning researchers and practitioners through our open-source repositories. 

New PhD students at MIT and other institutions are working on their thesis work building on top of our liquid neural network technology.  
 

Quickfire Questions

What's your power breakfast before a big day of innovation?

My juicy stack of new research works. 

The last book you've read that inspired an innovative idea or approach:

Build by Tony Fadell.

If you could master a new skill instantly, what would it be and why?
Getting back to playing piano! It really helps with some pauses in my heavily technical workflow.
 

What are your strategies for maintaining work-life balance and prioritizing self-care?

Spending time with non-technical friends, traveling, and hiking. 

If you could choose any innovator (real or fictional) as a mentor, who would it be and why?

Marc Andreessen.

Your favorite podcast that sparks creativity or offers insightful business trends:

“No Priors” with Elad Gil & Sarah Guo.

What is the most exciting innovation you've recently encountered in your field?

GPT-4 from OpenAI.

If you could invent any gadget to make your work life easier, what would it be?

Jarvis.

What achievement, opportunity, or aspect of your work are you most thankful for?

My team/students at MIT and my colleagues at my new venture as well as the quality of liquid foundation models we are developing.