The Twenty Minute VC (20VC): Foundation Models and the Future of AI with Rich Socher

Episode Title: "Foundation Models: Who Wins & Who Loses | How Economies and Labour Markets Need to Change in a World of AI | China vs the US in an AI Race: What You Need to Know | Rich Socher, Founder @ You.com"
Release Date: April 18, 2025
Host: Harry Stebbings
Guest: Rich Socher, Founder and CEO of You.com

1. Introduction

In this episode of The Twenty Minute VC (20VC), host Harry Stebbings engages in a comprehensive discussion with Rich Socher, founder and CEO of You.com. Rich brings an extensive background in artificial intelligence (AI), having served as Chief Scientist and EVP at Salesforce and CEO/CTO of the AI startup Metamind, which was acquired by Salesforce. His contributions to natural language processing (NLP) and foundational models position him as a leading voice in understanding the current AI landscape.

2. Current Landscape of Foundation Models

Rich Socher opens the conversation by addressing the rapid advancements and the overwhelming influx of updates in foundational models, particularly Large Language Models (LLMs). He describes AI as a "rising tide" with numerous "hype bubbles" emerging and dissipating. Rich emphasizes that while models like those from Sam Altman are likened to the capability of a "PhD student" ([06:11]), most everyday tasks don’t require such advanced intelligence. Instead, the true potential lies in integrating these models into businesses to enhance functionality and efficiency.

Notable Quote:
"AI is kind of this tide that's rising, but on top of that tide, you have a lot of little hype bubbles that come up and down." ([06:11] Rich Socher)

3. Commoditization of LLMs

The discussion shifts to the commoditization of LLMs. Rich agrees that LLMs are becoming commoditized, drawing an analogy to the telecommunications industry. He suggests that pure infrastructure layers of LLMs resemble telecom companies—high capital expenditure with significant value creation but limited value capture.

Notable Quote:
"LLM companies, especially just the pure thin infrastructure layer of LLMs are going to look, I think more and more like telcos in the sense that it's high capex huge expenditure to build it." ([08:13] Rich Socher)

4. Value Distribution and Moats in the LLM Space

Harry probes the sustainability of value distribution within the LLM ecosystem, highlighting the challenges of creating moats in a commoditized market. Rich acknowledges that consumer-focused companies like OpenAI, with their flagship product ChatGPT, hold significant market share, making other LLM applications appear as "rounding errors."

Notable Quote:
"If you're now just in that API infrastructure layer, it's very different. You have a lot more pressure." ([09:40] Rich Socher)

5. Specialization vs. Horizontal Agents

The conversation delves into the debate between specialized AI applications versus horizontal, general-purpose agents. Rich advocates for specialization, particularly in enterprise settings, where tailored AI solutions can address specific business needs more effectively than generalized models.

Notable Quote:
"We're seeing that play out in a variety of different ways. Like we're also focusing more on enterprise." ([16:28] Rich Socher)

6. Advertising in LLMs

Harry raises the topic of integrating advertisements into LLMs. Rich provides insights into why ads perform poorly in chat environments compared to traditional search ads. He underscores the importance of user experience, noting that intrusive or irrelevant ads can degrade the overall product quality.

Notable Quote:
"Ads work 10 to 100x worse than search ads in search." ([11:39] Rich Socher)

7. Future of LLM Adoption

Rich discusses the ongoing "unbundling wave" in consumer applications, where users prefer specialized apps for specific tasks over all-in-one solutions. He predicts that LLMs will gain traction in enterprise environments, where complex questions and efficiency are paramount.

Notable Quote:
"There's a huge unbundling wave. And so LLMs, as part of that unbundling wave of Google, LLMs will capture whenever you have more complex questions." ([13:07] Rich Socher)

8. Application of AI in Biology and Medicine

One of the most exciting applications of AI, according to Rich, lies in the fields of biology and medicine. He highlights how AI can unravel complex systems, such as simulating cellular interactions, which could lead to breakthroughs in understanding diseases and developing cures.

Notable Quote:
"AI is the perfect tool to tackle that kind of complexity because we can now have similar things, we understand how one neuron works, but when you have enough of them, you scale it up enough." ([25:19] Rich Socher)

9. Job Displacement and Social Systems

Addressing the societal impacts of AI, Rich expresses concern over immediate job displacement and the strain on social systems. He emphasizes the need for robust support mechanisms to help individuals transition into new roles as AI transforms various industries.

Notable Quote:
"Job changes are brutal in the moment. There will add a lot of pressure on social systems." ([34:06] Rich Socher)

10. Robotics and AI

The discussion touches on the potential of robotics to reach a "ChatGPT moment." Rich is skeptical, pointing out the challenges in developing humanoid robots that can handle the nuanced tasks humans perform daily. He argues that specialized robots may be more practical for specific industries.

Notable Quote:
"The tricky bit in robotics is that part of why ChatGPT had this amazing moment is that it's so general, right? You can just ask it anything." ([28:04] Rich Socher)

11. Quantum Computing and AI

Rich explores the intersection of quantum computing and AI, envisioning a future where quantum advancements enable the simulation of complex biological systems. While acknowledging the threat quantum computing poses to data security, he is optimistic about its potential to accelerate scientific discoveries.

Notable Quote:
"Simulate a cell and then multiple cells and organs and organisms, all of a sudden AI can try billions of different things on how to cure that cancer." ([50:07] Rich Socher)

12. Personal Reflections of Rich Socher

Towards the end of the episode, Rich shares personal insights and reflections. He discusses his evolving beliefs about AI timelines, the importance of programming skills, and his criteria for measuring success. Rich emphasizes positive impact over financial gain and expresses optimism about AI's role in advancing human knowledge and health.

Notable Quotes:

• "The biggest monarch is how much positive impact you've had in the world." ([55:35] Rich Socher)
• "AI will change in many ways. Science, I think, has been stuck in understanding the micro really, really well, but has not yet found a tool to understand complex systems really well." ([32:16] Rich Socher)

13. Conclusion

The episode concludes with Rich and Harry reflecting on the multifaceted implications of AI and foundational models. Rich underscores the necessity of strategic investment in AI, the importance of specialized applications, and the ethical considerations that come with technological advancements. His insights provide a nuanced perspective on the future trajectory of AI, emphasizing both its transformative potential and the challenges that lie ahead.

Key Takeaways:

• Commoditization of LLMs: While foundational models are becoming commoditized, value capture remains challenging, especially for pure infrastructure providers.

• Specialization Over Generalization: Specialized AI applications, particularly in enterprise settings, offer more sustainable value compared to horizontal, general-purpose agents.

• AI in Biology and Medicine: AI holds significant promise in unraveling complex biological systems, potentially leading to groundbreaking medical advancements.

• Societal Impacts: Immediate job displacement due to AI necessitates robust social support systems and rethinking of workforce training.

• Robotics and Quantum Computing: While robotics faces practical challenges in achieving general-purpose functionality, quantum computing could revolutionize AI-driven scientific research.

Rich Socher’s deep expertise and forward-thinking perspectives provide listeners with a comprehensive understanding of the current and future state of AI, highlighting both its transformative potential and the critical considerations required for its integration into society.