Professor McEthical- The First AI Ethics Copilot
Let us introduce you to Professor McEthical. The world's first AI Ethics Copilot, built with generative AI.
Earlier this year, Professor Christopher Sanchez was discussing with his team how they could help graduate students at the EGADE School of Business understand some of the highly complex issues in artificial intelligence, AI Ethics, generative AI, artificial general intelligence, labor impacts, and more. In the past, assigning massive amounts of reading would be the first step to get the students ready to tackle these issues. However, with the time constraints of the students and Professor Sanchez, we came up with the idea to develop an AI Ethics Copilot for the class that would cover all the essential and nuanced topics in AI.
We first searched to see if anything similar was online for commercial or academic use. To our surprise (and excitement), there wasn’t anything else remotely similar on the market in the startup, enterprise, or academic spaces.
Our first step was to build in guardrails, define the data sources, build the Fine-Tuners, and select the appropriate large language model to build on. Then we put Professor McEthical into the hands of some people we tasked with “red teaming” the system to ensure that the system would generate as few hallucinations and/or poor-quality content as possible.
The results were terrific. Not only did the system outperform all of the base scenarios we were testing. It could answer highly sophisticated and challenging questions put to it by us and others in our efforts to push the boundaries.
We then took things two steps further to create a truly useful system for the students, the general public, and society.
First, we developed a customized Professor McEthical, which was tailored with additional information for the class to serve as Co-Professor alongside Professor Sanchez. We are further developing this concept to deploy in academia to help professors amplify their presence, work, and interactions with students. This work is being done at Emergent AI.
We also trained several additional models based on the inputs from the students. These models were available for the students to work with during the course's final session. They tested the base system and the class system side-by-side to observe the impact (or lack of impact) their data had on creating a generative AI model. Students were also able to observe which student(s) and data set(s) were more relevant in generating the responses and uncovering how to conduct reinforcement learning with human feedback (RLHF.)
Second, we developed a LATAM AI Ethics Project that will grow over time and reflect the AI Ethics preferences of the region. The system will be available for governments, companies, and other parties to query when developing, procuring, designing, and deploying AI systems.
Third, we are creating a roadmap to create other similar models for other regions.
Interested in learning more, participating in the project, or something else? Reach out to us here.
*Professor McEthical was designed, developed, and funded by Professor Christopher Sanchez and his team at Emergent AI. The copilot is still in beta, may hallucinate, and should be fact-checked.
