- Lumian
- Posts
- Lumian Gen AI Newsletter Issue #56
Lumian Gen AI Newsletter Issue #56
DeepSeek>OpenAI?, Mistral’s Free Competitor, OpenAI’s Strategy
Welcome to the 56th edition of the Lumian Weekly Gen AI Newsletter!
First off, I’m excited to share that my mentor and Harvard Business School professor, Jeffrey Bussgang, is launching his new book The Experimentation Machine: Finding Product-Market Fit in the Age of AI! 🚀 The book reveals how founders can use AI as a superpower to become “10x Founders,” harnessing modern AI tools to run experiments more rapidly and efficiently.
Chock full of case studies from Bussgang’s popular HBS class (that I was fortunate to attend), in combination with modern AI approaches used by his portfolio companies (one of which I had the privilege of helping build), this book blends timeless startup wisdom with cutting-edge AI applications—a must-read for today’s founders.
Jeffrey’s knack for innovation doesn’t stop at startups. I had the incredible opportunity to collaborate with him on a really awesome project—ChatLTV, an AI-driven professor for Harvard. This first-of-its-kind initiative provides a glimpse into how AI could transform the future of education.
Now speaking of transformations and education, this week I want to cover Chegg’s story as it offers a cautionary tale of what happens when a business clings to an old model in a rapidly changing world…
Once, Chegg was the golden goose of college life. For $19.95 a month, students could access solutions to textbook problems. During the pandemic, as virtual learning soared, so did Chegg’s stock price, peaking at $15 billion. Professors grumbled about “Chegging” answers, but students swore by it. And then ChatGPT showed up, free, fast, and with no subscription.
Suddenly, “Chegg it” became a relic. Who needs to pay when a chatbot will solve your calculus problem and explain it with an enthusiasm only a robot can muster? Within months of ChatGPT’s debut, Chegg had lost half a million subscribers. Its stock, once flying high, plummeted 99%. The king of cramming was dethroned.
At first, Chegg’s leadership dismissed ChatGPT. Sure, it was impressive, but it wasn’t perfect—GPT occasionally made mistakes. Chegg banked on its army of contractors creating polished, human-verified solutions. Students, however, weren’t picky. They realized ChatGPT’s occasional errors didn’t matter as long as it got most things right—and for free. Why pay $19.95 for answers when you could have them instantly without even logging in? Chegg, the dependable old guard, suddenly looked like it was trying to sell CDs in the age of Spotify.
Realizing it was in trouble, Chegg partnered with OpenAI to launch Cheggmate, combining its vast database of answers with GPT-4’s AI magic. It sounded great on paper. In reality, it was more like a group project gone wrong. Students found Cheggmate’s answers underwhelming and its branding confusing.
By the time CEO Dan Rosensweig stepped down, the company was already in freefall. Nathan Schultz took over with plans to refocus on “curious learners” and expand internationally. But the realignment felt like rearranging deck chairs on a sinking ship. Meanwhile, students were still asking ChatGPT their homework questions, unfazed by its occasional errors because, hey, it was free.
Chegg didn’t throw in the towel entirely. Its homepage now features a textbox eerily similar to ChatGPT’s, asking, “What would you like help with today?” And AI automation has slashed production costs—an answer generated by AI is cheaper than one created by a human. But cheaper answers don’t matter much when students are still defecting en masse to ChatGPT.
Revenue is falling faster than a student’s GPA during finals, and Wall Street has all but given up hope. Chegg, once the darling of ed-tech, is now a cautionary tale: what happens when you cling to a business model while the ground shifts beneath your feet.
Chegg’s fate isn’t sealed, but its lesson is clear. The internet doesn’t forgive stagnation. ChatGPT didn’t just outpace Chegg; it rewrote the rules. Students used to “Chegg it.” Now they “ChatGPT it.” And if you’re curious what’s next for Chegg, just ask ChatGPT. It’s free—and it always has an answer.
Happy reading! 📚🤖🎵
In this week’s issue:
News Flash: DeepSeek>OpenAI?, Mistral’s Free Competitor, OpenAI’s Strategy
AI Frontier: AI Presentation Tools you can use today
Fundraising: The biggest deals in AI
Nerd Out: Technical and Business Content for Everyone
⏱️ News Flash
The 2-Minute Scoop to Keep You in the Loop
What's the Buzz?
Chinese AI companies like DeepSeek are now releasing open-source models that rival or surpass Western competitors, showing that innovation continues despite hardware restrictions.
Breaking It Down
DeepSeek's new R1-Lite-Preview matches the reasoning performance of OpenAI’s o1-preview on benchmarks like AIME and MATH, excelling in logic and transparency. Its ability to show real-time "thought processes" and improve over time marks a significant step, with open-source plans in the pipeline that could democratize access further.
Why It Matters
China's rapid progress in open-source AI challenges the dominance of U.S. tech giants like OpenAI, signaling a shift in the global AI landscape and intensifying the race for innovation and accessibility.
What's the Buzz?
French AI startup Mistral is shaking up the open-source AI landscape with cutting-edge updates to its chatbot Le Chat and the launch of powerful new AI models like Pixtral Large and Mistral Large 24.11.
Breaking It Down
Mistral’s Le Chat has added a canvas-like tool that makes it simple to create and edit documents, presentations, and even mockups in one place, perfect for brainstorming or polishing ideas. With new features like web search with citations, advanced analysis of PDFs and images, and automated workflows for repetitive tasks, it’s designed to help you work smarter and faster. Best of all, these updates are free during the beta, giving users a chance to explore powerful AI tools without a cost barrier.
Why It Matters
Mistral’s innovations signal an intensifying AI race, where open-source challengers are closing the gap with proprietary systems from U.S. tech giants like OpenAI and Anthropic, redefining global competition in advanced AI tools.
As this market heats up, the race to avoid commoditization lies in delivering faster, smarter, and more versatile models—anything less risks falling behind.
What's the Buzz?
AI companies like OpenAI are hitting a wall with traditional scaling methods, prompting new approaches to keep improving models like the upcoming Orion.
Breaking It Down
The leap in performance from one AI generation to the next is slowing as accessible training data runs out and hardware limitations rise. To address this, OpenAI and others are exploring techniques like test-time compute, which allows models to reason in real-time, and synthetic data to optimize performance without relying solely on scaling.
Why It Matters
The AI arms race is shifting gears—future breakthroughs may rely less on brute force and more on smarter strategies, potentially reshaping the balance of power in AI development.
🚀 AI in Practice
Cutting-Edge AI Presentation You Can Use Today
Gamma - AI powered storytelling
Beautiful.ai - AI presentation maker
🤑 Fundraising
The (AI) Intelligent Investor
🤖 Nerd Out
Technical and Business Readings
😜 Vote for the Future
Artificial General Intelligence (AGI) + Universal Basic Income (UBI)
How did you like this week's newsletter?Vote below: |
If you were forwarded this newsletter, you can access more of our content by subscribing here.
Best,
Reply