Sebastian Thrun (Stanford Adjunct Professor) – FICO World 2018 (Oct 2018)


Chapters

00:00:13 AI: From Physical Labor to Mental Agility
00:04:44 Neural Networks and Artificial Intelligence: Beating Humans at Their Own Games
00:10:52 AI-Powered Salesperson Advisors: Boosting Conversions and Tripling Sales
00:13:51 AI in Early Stage Cancer Detection
00:20:43 Artificial Intelligence and Deep Learning in Self-Driving Cars

Abstract

The Transformative Journey of Work: From Fields to AI Empowerment

The Evolution of Human Work

Human labor has undergone a remarkable transformation over the last three centuries. Originally rooted in agricultural fields requiring physical prowess, the advent of the steam engine catalyzed a shift towards increased productivity, allowing time for education and leisure. Today, 75% of the American workforce is stationed in offices, focusing on mental strength and routine tasks. Jobs such as lawyers, medical doctors, financial analysts, and spreadsheet operators involve routine and repetitive work.

AI’s Impact on the Workforce

Artificial Intelligence (AI) is redefining the nature of work by automating repetitive tasks. Its ability to infer rules from examples diminishes the need for complex programming skills, democratizing opportunities for those with less technical expertise. This paradigm shift is altering the skill set required in the modern workforce.

Deep Learning and Neural Networks

Deep learning, a branch of AI, uses neural networks to learn from data, showcasing its prowess in image recognition, language processing, and speech recognition. Neural networks are shallow mathematical simulations of the human brain with plasticity, allowing them to adapt and learn rules from data. They’ve been around since the 1930s, but recent advances in scale and data collection have made them more powerful. The essence of neural networks is their data-driven learning approach, eschewing the need for hard-coded rules. These networks, while not new, have become increasingly influential due to advancements in scale and data availability.

Scale: Unlocking Neural Network Potential

The exponential growth in data availability has been pivotal in enhancing neural network capabilities. With extensive datasets, these networks often surpass human learning capabilities, a key factor in their success in various applications, including FICO’s financial models.

Neural Network Successes and Challenges

Neural networks have triumphed in complex domains, such as games like chess and Go, and in automating tasks in sales and customer service. Deep Blue defeated Garry Kasparov in chess in 1997, demonstrating the power of massive compute power. In 2016, a deep learning program beat the world’s best Go player, Lee Vidal, through self-play and learning. AI systems are now beating humans in various professions, including sales and customer service. However, their ability to exhibit true intelligence and their societal impact remain debated and challenging areas of exploration.

AI-Enhanced Sales and Results

AI is revolutionizing sales by acting as an advisor to sales agents, using observed data to provide actionable suggestions. This collaboration has shown promising results, including significant increases in conversions and efficiency, as demonstrated by Cresta’s partnership with a Fortune 500 company. A Fortune 500 company Cresta equipped newly hired sales agents with an AI learning system as an advisor. AI-assisted agents achieved up to 22% more conversions compared to non-assisted agents. Over six months, the AI helped Cresta triple its sales conversions.

AI in High-Paying Professions

Highly skilled professions, such as dermatology, stand to benefit from AI’s capabilities. In skin cancer diagnosis, AI systems trained on extensive image databases have outperformed human dermatologists in accuracy, sensitivity, and specificity. This advancement showcases AI’s potential in enhancing healthcare delivery and accessibility. AI can assist doctors, particularly dermatologists, in diagnosing and treating skin cancers more effectively. AI can provide real-time guidance during surgery, helping surgeons make more precise incisions. AI can help doctors personalize treatment plans for patients, leading to better outcomes.

AI-Powered Skin Cancer Detection:

– Sebastian Thrun and his Stanford team created an AI system to detect skin cancer from images.

– The system was trained on a database of 129,000 images of various skin conditions, including melanomas and carcinomas.

– The AI system outperformed 21 board-certified dermatologists in identifying skin cancer, with a sensitivity error about a third of that of human doctors.

– The AI system is available on an iPhone app, making it accessible and affordable for many people.

Self-Driving Cars and Deep Learning:

– Sebastian Thrun, a notable AI expert, underscores deep learning’s role in developing self-driving cars.

– Udacity’s students, applying deep learning techniques, have achieved impressive results in autonomous driving, including a successful journey from Mountain View to San Francisco without human intervention.

– Thrun worked on developing self-driving cars, which have the potential to improve road safety and reduce traffic congestion.

– Self-driving cars use sensors, cameras, and AI to perceive their surroundings and make driving decisions.

– Thrun believes that self-driving cars will eventually become mainstream and revolutionize transportation.

Self-Driving Cars:

– Udacity teaches self-driving car technology to thousands of students, challenging them to apply deep learning to complex problems.

– Deep learning approaches have been used to develop impressive self-driving car capabilities, such as road surface detection and camera-based steering angle prediction.

– Udacity has successfully built a self-driving car that navigated from Mountain View to San Francisco using only cameras and radar.

AI’s Broad Impact Across Professions

Thrun highlights AI’s potential to augment human capabilities across various professions, from transportation to healthcare and business. He likens AI’s empowering role to past technological advancements, envisioning a future where AI liberates humans from mundane tasks to pursue more creative and fulfilling activities.

Conclusion

The journey of human labor, from manual fieldwork to AI-augmented professions, is a testament to our continuous evolution. AI, particularly in the fields of deep learning and neural networks, is not just altering the landscape of work but also enhancing human capabilities, offering unprecedented opportunities for efficiency, accuracy, and growth. As we embrace these changes, AI stands poised to redefine our professional and personal lives, ushering in a new era of human empowerment and technological synergy.


Notes by: Random Access