Key Moments
Why Everyone Is Wrong About AI (Including You) | Benedict Evans
The Knowledge ProjectKey Moments
AI is a platform shift like the iPhone, not an industrial revolution. Incumbents face disruption, and data isn't the sole advantage.
Key Insights
AI's current impact is akin to other major platform shifts (like the iPhone) rather than a more profound societal change like electricity or the industrial revolution.
Incumbent tech giants face significant disruption and re-evaluation of their business models during platform shifts, often struggling to adapt without cannibalizing existing revenue streams.
Proprietary data may not be as significant an advantage for AI training as widely believed, as the need for vast, generalized text data creates a more level playing field.
Predicting the winners and losers in the AI race is complex; past platform shifts show that early leaders can falter, and value capture can emerge in unexpected areas (e.g., search vs. browsers).
Regulation of AI is tricky; focusing on regulating a broad technology like 'AI' is less effective than regulating specific applications and understanding the trade-offs involved.
User adoption of AI chatbots is still nascent, with many users experimenting but not yet integrating them into daily, frequent workflows, suggesting a gap between potential and actual usage.
The true differentiator in AI may lie not in the underlying models but in productization, distribution, branding, and user experience, similar to the browser wars or social media platform success factors.
AI AS A PLATFORM SHIFT, NOT AN INDUSTRIAL REVOLUTION
Benedict Evans posits that AI represents a significant platform shift, comparable to the advent of the iPhone, rather than a transformative event on the scale of the industrial revolution or electricity. This perspective suggests that while AI will drive innovation and new applications for the next decade or so, its foundational impact is on how software and services are built and delivered, similar to previous technological paradigm changes. He cautions against overstating AI's immediate societal upheaval, framing it as the next logical step in computing's evolution.
THE CHALLENGE FOR INCUMBENTS IN DISRUPTIVE SHIFTS
Historically, platform shifts create moments of discontinuity where established companies are vulnerable. Incumbents often try to integrate new technologies as features or automate existing processes, which can be a strategy to avoid disrupting profitable core businesses. However, this can lead to them missing the fundamental changes. The automotive industry with electric vehicles and companies like Kodak with digital photography serve as examples where adapting to a new paradigm is fraught with existential challenges, often pitting the golden goose against the future.
DATA AS A LEVEL PLAYING FIELD
Contrary to the belief that proprietary data offers a decisive advantage in AI development, Evans suggests that the need for vast quantities of generalized text data levels the playing field. He argues that major tech companies' internal data reserves, while large, may not be significantly superior to publicly available data for training current large language models. This implies that the barriers to entry from a data perspective might be lower than anticipated, allowing for greater competition.
THE UNCERTAINTY OF VALUE CAPTURE AND COMPETITION
The true value capture in AI remains an open question. While many focus on the foundational models, the emergent value might lie in applications and user interfaces. The history of platform shifts, like the internet and mobile, shows that initial predictions about where value would accrue were often wrong. For instance, the early internet's promise was initially seen in areas like email, not the eventual dominance of search advertising and social media. Similarly, the smartphone revolution didn't accrue value for early mobile carriers but for device makers and app developers.
USER ADOPTION AND THE GAP BETWEEN POTENTIAL AND REALITY
Surveys indicate that while AI chatbots like ChatGPT are gaining traction, a significant portion of users interact with them sporadically or haven't fully integrated them into their daily routines. Many users experiment but don't find compelling, frequent use cases, leading to usage patterns of 'once a week' rather than daily engagement. This suggests that the 'black screen' interface of many AI tools requires users to actively identify needs, a barrier that might be overcome by AI integrated within more conventional products and workflows.
THE COMPLEXITY OF AI REGULATION
Regulating AI is viewed as an exercise in managing trade-offs, similar to regulating any other technology or economic activity. Evans likens it to regulating databases or spreadsheets, arguing that policy discussions framed around 'AI' as a monolithic entity are the wrong level of abstraction. Effective regulation necessitates understanding specific applications and considering the costs and consequences for product development, consumer benefit, competition, and national interests, rather than applying blanket restrictions that could stifle innovation.
THE FUTURE OF DISTRIBUTION AND BRANDING IN AI
In the current AI landscape, the underlying models are becoming increasingly commoditized, leading to a situation where differentiation might depend more on distribution, branding, and user experience. Just as browsers are functionally similar but differ greatly in market dominance due to brand and distribution (like Chrome), or social media platforms that offer similar photo-sharing capabilities but succeed based on network effects and user interface (like Instagram vs. Flickr), the future success of AI products may hinge on how they are packaged and presented to consumers.
ASSESSING CORPORATE POSITIONING IN THE AI RACE
Major tech companies are investing heavily in AI, with significant capital expenditure surges. Google is noted for making great models, though Meta's recent LLaMA 4 release faced criticism. Apple's strategy of 'doing it right' rather than being first, coupled with its integrated hardware and software ecosystem, positions it to potentially benefit from AI integration without being an early leader in AI models. Microsoft leverages its OpenAI partnership and Azure cloud infrastructure, while Amazon's AWS cloud business and advertising revenue provide strong foundations.
THE 'I-PHONIZATION' OF EXPERIENCES AND THE GOOGLE THREAT
A key concern is whether AI will lead to an 'i-phonization' of user experiences, where users access services through third-party models, potentially diminishing the direct engagement with platforms like Apple's iOS ecosystem. The threat to Google is that AI-powered search can become the new default, bypassing traditional search engine links and resetting user priorities. While Google has advantages, this discontinuity forces a re-evaluation of its entire product strategy and how search is monetized and delivered, especially as LLM chatbots become more capable.
THE ROLE OF AWARENESS AND INSIGHT IN LEARNING
Learning and pattern recognition are central to understanding complex fields like AI. Evans emphasizes that true insight comes from engaging with raw experiences and synthesizing information, not just consuming others' compressions. Similarly, in business, understanding what 'good' looks like in startups comes from exposure to both successes and failures. The ability to ask the next question, break down problems, and critically evaluate information is crucial, a skill honed through disciplines like history and philosophy, not solely through technical training.
DEFINING SUCCESS AND CURRICULUM IN A CHANGING WORLD
Success is personal and adaptable, often involving finding engaging work that meets financial needs and maintaining curiosity. Evans advises against rigid career paths, advocating instead for learning how to learn, think critically, and synthesize diverse information. In an era of rapid change, skills like adaptability, curiosity, and the ability to ask probing questions are paramount. The current tech landscape is moving so fast that understanding what is valuable and how to apply it is more important than mastering a single, potentially obsolete, skill.
Mentioned in This Episode
●Products
●Software & Apps
●Tools
●Companies
●Organizations
●Books
●Concepts
●People Referenced
Common Questions
Benedict Evans views AI as a major platform shift, comparable to the iPhone, which will drive innovation for the next 10-15 years, rather than an unprecedented, revolutionary force akin to electricity or the industrial revolution.
Topics
Mentioned in this video
Previously a disruptive area of focus, now seen as analogous to Japanese soldiers on isolated islands, unaware the war is over.
An early, centralized vision of what the internet would become.
A term used in 1995 to describe the interconnected computer network environment, before the internet's full potential was clear.
A book about Soviet central planning, used to illustrate the complexity that central planning cannot handle in sophisticated economies.
A book by a French academic that explores the concept of knowing about books without having read them and generalized pattern recognition.
A book by Zula about the creation of department stores, illustrating 19th-century innovation.
Mentioned as a book about financial bubbles, used to illustrate the recurring human tendency to believe current events are unique.
More from The Knowledge Project Podcast
View all 98 summaries
1 minWhy Customers Can't Figure Out What You Sell | April Dunford
2 minRobinhood CEO Calls Out the Banking Industry's "Stupid Tax"
2 min"They Called Us a Broken IPO" | Robinhood CEO
110 minVlad Tenev: GameStop, Founder Mode, AI
Found this useful? Build your knowledge library
Get AI-powered summaries of any YouTube video, podcast, or article in seconds. Save them to your personal pods and access them anytime.
Try Summify free