The Wharton School

I am a professor and Co-Director of the Generative AI Lab at Wharton, where I study and teach about innovation, AI, and entrepreneurship. I also build tools, games, and AI applications to democratize education.

Associate Professor

PhD Candidate in Management of Technology, Innovation, and Entrepreneurship

eMeta Corporation

VP Business Development/Co-founder

Mysterious and Exciting Past

Previous positions included co-founding a successful software company, inventing the transistor, squaring the circle, and winning the gold in the 1976 Olympics in gymnastics at the age of one.  I was also the original Lord of the Dance.

Various and Sundry

I research the factors that lead to success in entrepreneurship and innovation, and how to teach people to become more effective leaders and innovators. Most recently, I have focused on AI, and especially how it affects education and work. I co-founded Wharton Interactive, an effort to democratize education using games and simulations, while applying rigorous approaches to pedagogy at scale.

I have worked with a lot of different organizations (from startups to government agencies), co-founded a technology company, and have a mysterious past.

Google's NotebookLM is the current best "wow this is amazing and useful" demo of AI that most people will get immediately.

Here I gave it the entire text of my book (as well as my website and blog), and it turned it into a podcast, a study guide, FAQ, timeline & quite accurate chat that can reference parts of my entire book to support its points.

Listen to the first few minutes of the "podcast" it created for my book. Seriously, just listen.

(And yes, it is unnerving, too.)

Ethan Mollick

Associate Professor at The Wharton School. Author of Co-Intelligence

Avadhesh Singh

On a trip to Japan, ChatGPT advanced voice was incredibly useful.

Use custom instructions or say: “you are a translator between English and Japanese. You will wait for one person to finish talking and then translate into the other language. You will not add or say anything else”

Much more natural than other translation tools, and capable of handling nuances and cultural context.

The AI labs are great at building smarter and smarter AI systems. They aren't so great at understanding where smart AI systems can be useful.

A remarkably good investment, at the individual or firm level, is time actively experimenting to find use cases in your area of expertise. If you are waiting for use cases to be given to you, you are going to wait for all of your competitors to squeeze value out of current AI systems - and to be better positioned for the next generation of better AI systems.

AI is not being built for your use case. It is a general technology to adapt to your use case.

Bob Hutchins, MSc

A preview of the coming problem of working with AI when it starts to match or exceed human capability: Doctors were given cases to diagnose, with half getting GPT-4 access to help. The control group got 73% score in diagnostic accuracy & the GPT-4 group 77% in accuracy. No big difference.

But GPT-4 alone got 88%. The doctors didn't change their opinions when working with AI.

To be clear, this doesn't say AI will always beat doctors - this is a narrow test. It is much more about what this means for the future. As AI models get better, and match or exceed human level performance, what happens? This is one example where it is happening, and we see the issues emerging.

Note: I made a couple edits to this post 2 hours after posting to clarify issues raised in the comments. Specifically to make it clear that we do not have a reason why doctors did not benefit from AI, and to clarify this was diagnostic accuracy and not final diagnosis rates.

Paper: Paper: https://lnkd.in/eCYhh-a2

Not what massaging the data means.

(I have been randomly animating scientific diagrams with Pika)

This August 2024 US survey of AI use confirms that AI is not just hype:
1) AI adoption is crazy fast by historical standards
2) It obviously useful to people in many industries. 25% of people use GenAI at least 60 minutes a day on days they use it for work, already
3) AI is being used for many purposes

We can stop debating “does anyone find AI useful” on this site, at least. Paper: https://lnkd.in/gyds7UKi

No alternative text description for this image

Justin Lahullier

This (from someone at OpenAI) is an important thing to understand

The major AI labs, including OpenAI, are very much focused on racing for the future. It is almost accidental that their early products are making billions of dollars. Their goal
Is explicitly AGI, not your use case. They are not spending much time commercializing (yet) because most of their effort is going to building more powerful AIs.

AI can help learning... when it isn't a crutch.

There are now multiple controlled experiments showing that students who use AI to get answers to problems hurts learning (even though they think they are learning), but that students who use AI as a tutor perform better on tests.

Papers:
https://lnkd.in/eXF4uP95
https://lnkd.in/efRCTUdC
https://lnkd.in/d_xS7XWm

You really, really should not trust audio clips anymore.

Even a couple months ago, it used to take a commercial service to clone a voice. No more. Here is me creating a voice clone of myself using just a 10 second reference clip on my home computer using open software.

This is all real time, no cuts.

At OpenAI's education event last night, heard from a couple professors who were replacing purchased class readings with AI-generated (importantly, vetted/edited by the professor!) customized readings that synthesized a lot of content and were designed to better fit the syllabus.

Given this is going viral, two points: (1) there are obviously lots of exciting and troubling implications to this, but it is not a theoretical discussion, it is something that is happening in the world, and (2) to be clear, no one (I heard from) was doing this for literature or other assignments involving primary sources or where the readings were subject to discussions.

The difference between ChatGPT Advanced Voice & Google and Meta’s AI voice is Advanced Voice is multimodal - the model “hears” your voice, emotions & tone. It can make any sound or voice in reply (this is limited by OpenAI right now). Other models are text-to-speech. The AI’s text is simply read to you, and your replies are turned into text to feed the AI.

Native multimodality is extremely powerful in ways that many people haven’t fully understood yet. A more multimodal model can produce words, or images in roughly the same way it makes text. This gives it much more direct senses of the world, and fine-grained control over outputs, expanding its capabilities.

Google & Meta likely have complete voice multimodal models. They just haven’t released them yet.

Fascinating (premature?) victory lap from Sam Altman & OpenAI. This is quite the declaration.

Take this sort of stuff with a grain of salt, but also as useful signal about attitudes of AI insiders actually building new models. https://ia.samaltman.com/

Taplio

Ethan Mollick's Linkedin Analytics

's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Ethan Mollick's Linkedin Analytics

's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Ethan Mollick's Linkedin Analytics

's Best Posts (last 30 days)