Innovation Profs - 3/14/2025

Your weekly guide to getting the most out of generative AI tools

Welcome to our Friday edition of the Innovation Profs newsletter. Look for us in your inbox every Tuesday with our usual news-filled newsletter and on Fridays to focus on making you more productive with generative AI. Let’s get started…

Why AI ethics?

In our presentations to a wide range of audiences, the most frequent category of questions we encounter involve the ethics of generative AI. These questions often come in two flavors: The first is, “What are some of the ethical concerns raised by generative AI?” and the second is, “How can I use generative AI responsibly?”

In our view, these questions can be seen as two sides of the same coin, as many of the ethical concerns raised by generative AI determine what counts as responsible use of generative AI. In other words, to be a responsible user of generative AI, you need to know what ethical issues might arise in using generative AI and how to respond to those issues.

Here’s how we’re explaining this to our students. This semester, in our course on generative AI, for each of the category of AI generation tools we’re covering in class — text, image, audio, video — we discuss ethical and legal issues raised by these technologies, framed by what we refer to as our “ethical cast of characters,” depicted below (as a throwback to Porter’s childhood, these are conceptualized as a sequence of bosses you have to defeat in a Mega Man-style game):

Let’s introduce a few of these characters.

Hallucination ghost

Hallucination

First up is Hallucination, representing the well-known fact that large language models can yield incorrect outputs. If you open up ChatGPT in your browser, you’ll notice the disclaimer beneath the input text that states, “ChatGPT can make mistakes. Check important info.” Immediately, we should ask: Why do these mistakes happen? Why would anyone use a tool that can make mistakes? If we do decide to use tools like ChatGPT, how can we use them in a way that minimizes the possibility of making mistakes?

Bias

Next, we have Bias, which represents the fact that generative AI tools can propagate any biases that are explicitly or implicitly contained in the data used to train the models underlying these tools. For instance, when we prompt an AI image generation tool to depict a person performing a certain occupation, it would be problematic if every instance produced were only of one specific category of individuals. Again, we might ask some questions: What steps are being taken to avoid this type of problem? How do I ensure that I am mitigating bias in my own use of generative AI?

Security and Privacy

Finally, a third character represents the dual nature of Security and Privacy. This is a common concern we are regularly asked about. What happens to the data we enter into LLMs? Is there any way for LLMs to handle proprietary information? How do we protect our customer’s privacy if we use LLMs to process their data?

What all of the questions raised above have in common is that we need to know how to answer them if we want to be responsible users of generative AI.

Want to learn about general strategies for navigating these issues? Eager to hear about the other members of the ethical cast of characters? Sign up for our next AI Lunch Club: AI Ethics on March 26. You’ll learn about…

  • The main ethical issues raised by generative AI, including the source of these issues

  • General strategies for navigating these issues

  • A general framework for gauging responsible use of AI on a case-by-case basis

  • And more!

Sign up for more Lunch Clubs:

  • April 9: AI in Law and Policy The aim of this session is to give a broad overview of the issues, informed by the latest developments at the intersection of AI and law/policy.

Generative AI Mailbag

This week’s question: What is the best generative AI tool to use for writing and brainstorming?

Answer: It does seem that ChatGPT is winning this race so far. But there are plenty of other options in this area, including Google Gemini, Anthropic’s Claude, Microsoft Copilot, Meta’s Llama and xAI’s Grok.

The best advice we can give you right now is to try them all to see which one works best for your needs. You can also check out the Chatbot Arena Leaderboard to see which tools are scoring the best currently. Or join the fun and test tools in the Chatbot Arena.

Do you have a question about generative AI? Submit it by replying to this email or sending an email to [email protected].