PodcastsThe Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch20VC: Does Value Accrue to Incumbents or Startups in the AI Race, Why Model Size Matters More Than Data Size, Why Artificial General Intelligence is Far Away, Why Carpenters Will Be Paid More Than Software Engineers & Future of Jobs with Richard Socher
20VC: Does Value Accrue to Incumbents or Startups in the AI Race, Why Model Size Matters More Than Data Size, Why Artificial General Intelligence is Far Away, Why Carpenters Will Be Paid More Than Software Engineers & Future of Jobs with Richard Socher
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The PitchThu Aug 17 2023
Richard Socher's Journey into AI:
- Richard started his journey in AI over a decade ago, studying linguistic computer science at Leipzig University.
- He then switched to computer vision for his master's degree and focused on statistical learning and pattern recognition.
- During his PhD, he became interested in deep learning and neural networks for natural language processing, which led him to invent contextual vectors and prompt engineering.
The Landscape Evaluation of AI:
- There is a current hype cycle around AI, but it is important to evaluate the historical context of AI development.
- Richard believes that there has been significant progress in AI capabilities, but there are also inflated expectations about continuous exponential improvement.
- People often think that every task can be automated with AI, but there are still limitations and misconceptions about what AI can do.
Importance of Model Size in AI:
- Model size is crucial in training models for different tasks in natural language processing.
- Large models are necessary to train a single model for all NLP tasks, as smaller models would not have enough parameters to learn complex predictive functions.
- Data size is also important, but it is not mutually exclusive from model size. Both large models and large amounts of training data are needed for effective performance.
Value Accrual: Startups vs Incumbents:
- The value accrual in the field of AI can happen both for startups and incumbents.
- Startups have the advantage of being able to quickly innovate and adapt to new technologies, while incumbents have access to customer data and resources.
- It is difficult to predict which incumbents will succeed or lag behind in the race for value accrual in the field of AI.
Open vs Closed Ecosystems:
- There is an ongoing debate between open and closed ecosystems in the field of AI.
- Open ecosystems allow collaboration and contribution from various researchers and developers, leading to faster innovation.
- However, closed ecosystems provide advantages such as proprietary data and resources, which can lead to faster progress for specific companies.
Carpenters vs Software Engineers:
- Richard believes that in the future, carpenters will be paid more than software engineers.
- This is because physical tasks are becoming more expensive as AI automates many digital tasks.
- The value of manual labor and craftsmanship will increase as it becomes scarcer compared to automated digital work.
AGI and its Challenges:
- AGI (Artificial General Intelligence) is often overhyped and misunderstood.
- Richard believes that there are still significant research breakthroughs needed to achieve AGI.
- The concept of AGI having its own goals and intentions is a challenge, as current AI models are designed to fulfill human-defined objectives.
- The development of AGI requires advancements in understanding intelligence beyond token prediction.
Transition Period in Technological Revolutions:
- Richard acknowledges the concerns about job displacement during technological revolutions.
- He emphasizes the importance of supporting individuals through transitions by providing training and assistance in adapting to new technologies.
- While some jobs may be automated, new opportunities and industries can emerge from technological advancements.
The Future Impact of AI:
- Richard predicts that AI will play an even bigger role in society in ten years' time.
- It will continue to improve efficiency, provide better answers, and transform various industries.