Sign In

Steven Rick

Postdoctoral Research Associate in the Center for Collective Intelligence

Steven Rick is a Postdoctoral Research Associate at the Massachusetts Institute of Technology (MIT). He is described as a full-stack researcher and developer, indicating a strong background in both research and practical application of technology. His educational background includes experience at the University of California San Diego.

On LinkedIn, he actively shares insights related to his research, including topics on human-AI collaboration, as evidenced by his recent posts about published research on scaffolding such collaborations.1

Highlights

Jul 19 · linkedin

🚀 Excited to share that our latest research is now published online! 🚀

"Supermind Ideator: How Scaffolding Human-AI Collaboration Can Increase Creativity" - https://lnkd.in/ecaqsnq4

Our team wanted to help people think about ways to solve problems through collective intelligence. To do this in a scalable manner we built a platform - Supermind Ideator - which took the methods from Supermind Design and combined them with Large Language Model prompting and fine-tuning to produce behavior that is similar to that of a workshop facilitator.

We liked what we saw but wanted to understand if this constrained approach to LLM interaction was actually helping or hindering open-ended creative problem-solving. So we ran a study! 152 people created 611 solutions to two counterbalanced problem statements. They either worked alone, used ChatGPT, or used our Supermind Ideator. 505 new people then evaluated those 611 ideas.

🔍 Key findings 🔍

  • Those with access to 'external' help (ChatGPT, Ideator) tended to produce solutions that were considered more innovative than those who worked alone.
  • Those who used Supermind Ideator tended to see more ideas and produce solutions that were more innovative than those who used ChatGPT.
  • There was no difference in the number of solutions submitted across conditions.
  • People tended to engage minimally with both Supermind Ideator and ChatGPT, taking the least 'turns' to accomplish the task

🧠 My interpretation 🧠 LLMs (and many other recent advances in AI) can help people explore problems and generate many potential solutions, but it requires know-how in order to really maximize value from the machine's output. Left to generic prompting, LLMs will 'satisfice' the end user quickly and lose much of the power of critical thought. Our Supermind Ideator imposed many more constraints on interaction than ChatGPT did, and that proved to be a good thing. I believe this stems from how our UI and API scaffold out LLM interactions on behalf of the end user, empowering them to ideate differently than they would have on their own or with a ChatGPT-like interface. We're still trying to solve that 'satificing' issue - but we think we're on to something and excited to share.

This work is far from finished. We are actively working on a new version of the platform which we believe will further enhance human-machine creative problem-solving.

The Supermind Ideator is live here: https://ideator.mit.edu/ New users can request access via the waitlist.

Jennifer Heyman Gianni Giacomelli Haoran Wen Robert Laubacher Nancy Taubenslag Max Sina Knicker Younes J. Pranav Ragupathy Jared Curhan Thomas Malone

Feb 15 · linkedin

The latest video from Kurzgesagt struck me in a surprisingly personal way. I remember sitting in highschool science when the reality of the impossible machine that is the human body fully hit me. All the chemistry, biology, and physics suddenly clicked together and I became enamored with the way science tries to take wildly complex things and break them down into their consistituent parts (thanks electron orbitals, Krebs cycle, and electromagnetic forces).

That spark pushed me to study Cognitive Science as an undergrad because I wanted to better understand the interplay of simple physiology that work together to drive complex human perceptions and experiences. A doctorate in Computer Science took that further to understand the parts and pieces of computing that combine to manifest in complex sociotechnical systems in the world today.

I’m now working on emergence-adjacent topics, trying to find new and powerful ways to combine the capabilities and expertise of machines and humans to create more effective and powerful teams.

I’m sure there has been a lot of good fortune to bring me here, but looking back and seeing the connectivity continues to surprise and further inspires me to teach students to stay curious and chase those sparks. See the simple pieces that combine to build incredibly impossible machinces.

#science #cognitivescience #computerscience #curiosity #collectiveintelligence #emergence

Related Questions

What are Steven Rick's main research interests at MIT?
What projects has Steven Rick worked on at MIT?
What is Steven Rick's educational background?
What are some notable achievements of Steven Rick at MIT?
What is Steven Rick's role at TruStage?
Steven Rick
Steven Rick, photo 1
Steven Rick, photo 2
Add to my network

Location

Greater Boston