Aha Moment:

Eric Schmidt identifies his first “aha moment” with AI as occurring in 2011 when a group experimented with YouTube data. The experiment revealed the concept of cats through unsupervised learning, an initial discovery that opened up new horizons.

Birth of Google Brain:

Following this discovery, Google X started the Google Brain project. This initiative led to significant advancements in language models, such as BERT and Transformers. These advancements have had a transformative effect on Google’s search and advertising sectors, driving gains in revenue and search quality.

Acquisition of DeepMind:

Eric mentions the acquisition of DeepMind as another significant milestone. Initially considered expensive, the purchase is now seen as a smart investment, given DeepMind’s global leadership in reinforcement learning.

AI as a Business Catalyst:

Schmidt describes AI companies as those that involve some form of learning. In consumer tech, more clicks translate to more learning opportunities. This enhances both the quality of services and the targeted advertising, which in turn drives customer satisfaction and revenue.

Traditional vs AI Teams:

Schmidt highlights how AI and traditional algorithmic approaches were pitted against each other in Google. Interestingly, while the outcomes were often similar, the routes taken by AI were different, revealing correlations and patterns that humans could not see. This has led to significant gains in various sectors within Google.

Generative Models:

Schmidt discusses the current focus on generative models in AI, which have capabilities beyond pattern recognition. These models can predict likely outcomes, a powerful tool that is driving enthusiasm among consumer tech companies.

Impact on Business Models:

AI directly improves customer quality and revenue. It allows for individual targeting without requiring profiling, although Schmidt acknowledges that the technology is not yet perfect. The incorporation of AI is driving significant changes across business models.

Global Rivalry in AI:

Eric Schmidt states that the central global competition regarding AI is not between companies like Google and Apple, but rather between countries, specifically the United States and China. The U.S. has built an innovation model that has led to advancements such as GPS, semiconductors, and the internet. China, however, follows a different but equally powerful model.

China’s Strengths:

China employs a contrasting yet formidable approach involving a massive pool of engineers, robust work ethic, and a vast scale of platforms. Companies there often follow a 9-9-6 work schedule (9 a.m. to 9 p.m., six days a week), which Schmidt notes is illegal in the U.S.

Internet’s Evolution:

The internet has evolved from being optional to fundamental, impacting every form of human activity. This fundamental nature of the internet affects how governments shape their geopolitical strategies and internal policies.

Internet Splintering:

Schmidt predicts a splintering of the internet, particularly at the application layer, due to differing national interests and policies. China and the U.S. are each large enough to create their “own weather,” affecting global internet structure and dynamics.

Regulatory Approaches:

China has tackled internet governance by making anonymity online illegal, enabling tighter control over its citizens. This contrasts with the more libertarian internet culture in the U.S. Schmidt emphasizes that each country’s approach aims to create a stable information space tailored to their governance needs.

Outcome Prediction:

Schmidt believes that China’s strategy to maintain power and control through technology will likely succeed. He makes it clear he is not endorsing China’s approach, but merely stating that they have a well-defined strategy backed by sufficient resources.

Technology Industry’s Unique Culture:

Schmidt also touches upon the unique sociology within the tech industry, emphasizing that the worldview in this sector differs substantially from broader society. This distinction is significant because the tech industry’s mindset impacts how AI and internet technologies evolve globally.

In summary, Schmidt underscores the geopolitical dimensions of AI, positioning the U.S. and China as the central players whose differing strategies and approaches will have global ramifications.

Regulatory Approaches:

Eric Schmidt highlights how China takes a direct regulatory approach towards the internet, which includes using machine learning to find and block VPNs. The country is unambiguous about its intent to govern the digital space, and this separates it from the more laissez-faire strategies found in Western democracies.

Global Alignments:

Schmidt discusses the potential alignments of other nations with either China or the United States in their approach to the internet. Democratic nations are more likely to side with the U.S., while authoritarian regimes or weaker states will likely be influenced by Chinese architecture. The reason being, they can benefit from China’s infrastructure and technology.

Trade Influence:

Countries economically dependent on China may find themselves in a difficult situation. Schmidt uses Australia as an example. Australia is economically reliant on China, its largest trading partner, but has taken a firm stand against China on certain ideological fronts.

Belt and Road Initiative:

Schmidt notes that countries involved in China’s Belt and Road Initiative are likely to adopt Chinese information rules, including surveillance and censorship. This is because these countries will get their technology infrastructure from China.

The European Dilemma:

European nations like Germany find themselves in a unique quandary. They are economically aligned with China, but ideologically closer to Western norms. Schmidt predicts a form of “uncomfortable competition,” where countries will have to make tough decisions about which world view to align with.

Case of Smaller Nations:

Schmidt discusses Tunisia as an example of a country that could face a dilemma. It’s a well-run nation compared to its peers, but it would have more financial opportunities with China than the U.S. The question becomes how countries like Tunisia will decide which path to take in this dichotomy.

AI’s Dual Nature:

The dialogue starts by addressing two approaches to AI development: one that focuses on data centralization and is more authoritarian in nature, and another that prioritizes ethics, privacy, and avoiding bias. These approaches indicate a global tension, especially as China leads in setting global AI norms.

Importance of Algorithms:

Eric Schmidt highlights that while China may have an edge in terms of data and lax privacy regulations, algorithms are just as crucial. He refers to a commission he led, recommending doubling research funding and building a national research network to level the playing field.

Enhancing Western Practices:

Schmidt stresses that the West can’t merely adopt someone else’s good practices but must strengthen its own. This involves welcoming high-skill immigration, increasing research funding, and infusing technical talent into governmental bodies.

Governmental Technical Gap:

The American government, according to Schmidt, is largely unequipped to even understand the complexities of AI. Most AI work in governmental departments has been done by volunteers and consultants. This represents a vulnerability.

National Security Challenge:

Schmidt views AI as a national security challenge for the United States. He emphasizes the need for collective effort to ensure American technology, values, and startups remain global platforms. China is already a major competitor in various fields, including AI, and is rapidly catching up.

Government Preparedness:

The government needs to prepare for how digital technologies and AI will alter systems fundamentally. Key initiatives should go beyond fostering innovation to include acquiring technical talent capable of anticipating systemic changes due to emerging technologies.

Overall, the conversation underscores the urgency for Western nations, particularly the United States, to focus not just on AI’s technological aspects but also on ethical considerations, government preparedness, and international competition, especially with China.

Private Sector’s Role in Innovation:

Eric Schmidt emphasizes that most technological innovation today occurs in private companies, a shift from 50 years ago when government labs and universities were the primary sources. He argues against premature regulation, suggesting that it could stifle innovation. Schmidt contrasts this with China, which is more focused on starting new initiatives rather than halting them due to regulation.

National Strategy for Security:

Schmidt believes a coherent national plan is essential, especially when national security is at stake. He cites the example of Operation Warp Speed, a collaborative effort involving universities, the private sector, and government, as an effective model for handling national emergencies through innovation.

Importance of Top-Down Leadership:

He says that if you want to change a company, you can do it from the bottom up through customers, but changing government requires a top-down approach. Current systems, especially in the Defense Department, are not built for rapid change. Timelines for implementing recommendations in government can span years, unlike in agile private companies.

AI’s Transformative Potential:

Schmidt states that AI will be a transformative force in many sectors, from healthcare to material science. He predicts that the next trillion-dollar industries will arise from the application of digital technology and AI in sectors like drug discovery and healthcare, which comprise a significant portion of GDP.

Urgency and Competitiveness:

The conversation notes that governmental timelines for implementing changes are often long, raising questions about the urgency needed to remain competitive, especially against fast-moving competitors like China. Schmidt implies that the window for effective action may be limited, emphasizing the need for quick, decisive action.

TikTok as a Wake-up Call:

Eric Schmidt uses TikTok as an example of China’s rapid technological advancement. Despite efforts by President Trump to make TikTok a U.S. operation, it remains Chinese. Schmidt attributes TikTok’s success to its innovative AI algorithm, which matches users based on interests rather than social connections.

Timeframe for Action:

Schmidt asserts that the U.S. has a short window, maybe one to two years, to organize around technological initiatives. He emphasizes the need for more research funding, a national research network, and working with allies.

Consistent Guidelines and Ethics:

Schmidt highlights the importance of establishing guidelines and ethics that align with American values. He stresses that specific recommendations have been made for how the Defense Department and intelligence communities should operate, generally in the form of elevating certain functions and resourcing them adequately.

Education and Talent Utilization:

He proposes the creation of a civilian university for technical talent, which would be free in return for up to five years of government service. Additionally, Schmidt suggests a reserve corps modeled after ROTC, allowing professionals to contribute to the government on a short-term basis.

Public Willingness to Contribute:

Schmidt believes there is significant public interest in aiding the government to reach optimal outcomes in terms of national security and ethics. He concludes that, with the right structure and urgency, these goals are achievable.

Overall Urgency:

Both speakers agree on the sense of urgency required. The time for action is measured in a few years, not decades, emphasizing the need for quick and effective initiatives.

AI’s Balance of Power:

Eric Schmidt discusses the balance of power between the U.S. and China in terms of AI, particularly in cyber warfare. He says that an AI-based offensive or defensive cyber weapon could tip the scales. Unlike other technologies that are easier to contain, AI knowledge leaks rapidly, making it difficult to maintain a long-term advantage.

Uncontainable Nature of AI:

He points out that AI is fundamentally different from previous technologies, like nuclear capabilities, in that it cannot easily be contained. This poses new challenges for national security, as the technology can quickly be transferred and replicated, thereby destabilizing traditional paradigms of power balance.

AGI’s Game-Changing Potential:

Schmidt elaborates on Artificial General Intelligence (AGI), suggesting that if one country develops a highly advanced or dangerous system, it could create a new level of existential threat. He posits that regulatory frameworks similar to nuclear non-proliferation treaties may become necessary for AGI.

Need for New Governance:

Schmidt highlights that the increasing governmental use of cyber technology for influence, as seen in the Russian interference in 2016, is a new phenomenon. The discourse lacks a language to discuss how to maintain balance of power or contain dangerous AI technologies, something that urgently needs to be addressed.

AI’s Broad Applicability:

Schmidt states that AI has the power to transform almost all industries, making it a foundational technology unlike any other. Its implications are broad and can even be applied in synthetic biology to create new drugs or treatments, exemplified by the antibiotic Halicin developed at MIT.

Societal and Ethical Concerns:

Schmidt warns that AI’s power in information spaces can be exploited for spreading misinformation. As AI gets better at understanding human biases, the potential for abuse increases. He even delves into how AI could become a part of a child’s upbringing, posing ethical dilemmas around governance and behavior.

AI in Healthcare and More:

Lastly, Schmidt talks about the potential for AI in healthcare, citing the ability of AI to predict why a patient might be showing up at a hospital, amongst its other applications. This transformative nature of AI can lead to breakthroughs in various sectors, from inventory management to synthetic biology.

Eric Schmidt’s dialogue reveals the complex landscape of AI, touching upon the critical challenges it presents in national security, governance, and ethics, while also highlighting its transformative capabilities across various industries.

Genetic Similarities and AI in Healthcare:

Eric Schmidt discusses the potential of AI in healthcare by highlighting how it can find genetic similarities among people and diagnose diseases more efficiently. AI can analyze vast data sets to find patterns, thereby helping doctors in their diagnostic process.

Scientific and Economic Implications:

Schmidt talks about AI’s role in scientific discovery, stating that it enables researchers to conduct more science digitally. This reduces the need for costly lab work, consequently making scientific processes more efficient and less expensive.

Scientific Renaissance:

He suggests that AI could lead to a renaissance in scientific discovery. As an example, he mentions a Caltech project he funded that aims to develop a new forecasting model for climate change, focusing on cloud behavior. Traditional simulation models like Navier-Stokes equations are not sufficient to simulate complex phenomena like clouds due to computational limitations. However, AI can provide approximations that are workable.

AI in Unconventional Fields:

Schmidt recounts an example where AI was used to determine if a mouse was asleep. Lacking labeled data, scientists generated synthetic training data to build an accurate model. This exemplifies how AI can be used even when traditional scientific data is insufficient.

Two Facets of AI in Science:

He identifies two primary uses of AI in science. The first is approximating functions that are otherwise computationally infeasible. The second is the generative aspect where AI can produce new things for testing, advancing the scientific exploration process.

Role in Climate Change Innovation:

Schmidt asserts that AI will likely be pivotal in addressing climate change through innovation. Traditional approaches to solving climate change have not been efficient enough, and AI’s capacity to handle large data sets and provide approximations could lead to significant breakthroughs.

Economic Impact of AI on Industries:

Eric Schmidt outlines the vast potential for AI to disrupt various industries, especially those that are lightly regulated. He points out that companies who are quick to adopt AI can significantly outperform their competition, which could result in the loss of jobs and shareholder wealth.

Job Quality and Income Disparity:

Schmidt touches on the societal implications of AI adoption, specifically mentioning the potential for fewer high-quality jobs. Winners in the AI transformation are likely to hoard more resources, exacerbating the Gini coefficient or income disparity.

AI for Social Good:

Schmidt advocates for the use of AI to address societal issues, such as the justice system’s inefficiencies. He offers an example of AI being used to analyze the behavior of parole officers to determine inconsistencies in parole violations.

Cognitive Work vs. Manual Labor:

Schmidt comments on the uneven advancement in AI capabilities, noting that cognitive tasks are being automated at a faster rate compared to manual tasks, which have been harder to automate. This is attributed to the digital nature of cognitive work.

Rate of Technological Change:

Schmidt predicts that the next 10 to 20 years will see an accelerated rate of technological change compared to the past decade. This rapid change could outstrip humanity’s ability to adapt and could bring about other problems and disparities.

Regulated vs. Unregulated Industries:

Schmidt draws a distinction between industries that are regulated and those that are not. Regulated industries are less likely to see rapid AI adoption because of existing agreements between regulators and companies.

The Future of Education:

One industry Schmidt focuses on for AI application is education. Despite the sector taking up a significant portion of GDP, there has been little innovation in optimizing learning. He urges for corrective action to harness AI for better educational outcomes.

Business Physics Changed by Software and AI:

Eric Schmidt discusses the transformative impact of software and AI on business. In the early stages, software introduced the concept of “free replication,” which fundamentally changed how businesses operate. The advent of AI and machine learning has introduced a “new physics” into business ecosystems, characterized by systems that improve rapidly and offer compounding gains in quality, effectiveness, and functionality.

Concentration of Power and Vision for Technology:

The discussion shifts to the concentration of power in countries like China and leading companies in the U.S., a reality that diverges from technology’s original vision rooted in decentralization and individual freedom. The existing concentration of economic, social, and moral power might not be a permanent state; technology could oscillate between centralized and decentralized structures.

Futures Unknown:

The conversation explores the unknown impacts of future technologies. There’s speculation about various scenarios where the next generation of companies could either become immensely powerful due to specialization, or no single company would dominate because of market diversity.

Business Models in the Age of AI:

When discussing potential future business models enabled by AI, Schmidt emphasizes the importance of building a product that is 10 times better than existing solutions. He suggests that if a business can attract users and grow, making money will not be an issue. Various monetization avenues such as licensing, selling, and transferring technology can be explored.

Build a Great Product:

Schmidt’s closing thoughts focus on the value of creating a superior product. Rather than worrying about perfecting a revenue model from the start, Schmidt advises focusing on product quality to attract users. Once there is a user base, the pathways for revenue generation become more accessible.

The dialogue aims to offer a broad perspective on the technological shifts affecting business paradigms, touching on decentralization, the evolution of business models, and the pivotal role of product quality.

Grand Challenges in AI:

Eric Schmidt discusses the grand challenges in AI beyond the long-term goal of achieving Artificial General Intelligence (AGI). He suggests that more immediate challenges could be exciting and deeply impactful, like AI’s role in combating climate change.

Evolution of AI Systems:

Schmidt notes that AI technology has advanced from basic self-supervised and partially supervised learning to multi-model reasoning systems. These systems generate candidates and refine them through complex two or three-stage pipelines. The construction of these systems requires deep expertise, often at the PhD level.

Accessibility of AI:

Schmidt argues for making AI technologies more accessible to individuals with a “normal technical education.” He points out that initially, many technologies, like email, were complex and niche but eventually became universally accessible. Schmidt wants AI to follow a similar trajectory.

The Role of Education:

Schmidt emphasizes the importance of not just focusing on top-tier universities for AI expertise but also large state schools that produce the bulk of technical talent. He suggests that the tools should be designed so that these students can tackle complex tasks without necessarily understanding the deep intricacies.

Importance of Open Source:

Schmidt credits the open-source movement for accelerating knowledge sharing and innovation. He advocates for collaborative platforms that can expedite scientific discoveries, enabling the next generation to solve critical problems more effectively.

Call to Action:

The conversation ends with a call to democratize AI and machine learning, making it accessible to a broader range of people, from industry specialists to those with a general interest in technology.