One of the most mortifying aspects of knowing many tech enthusiasts is listening to them excitedly describe what they believe are groundbreaking discoveries. Recently, an acquaintance enthusiastically explained to me that knowledge is structured into language, as revealed by large language models (LLMs). He claimed that putting a word into ChatGPT could demonstrate understanding, and that LLMs are a discovery on par with writing. This idea, however, is a naive version of Structuralism, a concept from a century ago. I quickly exited the conversation, noting his frustration at my lack of agreement—a behavior possibly symptomatic of LLM overuse.
The Hubris of Tech Enthusiasts
Not every discovery new to an individual is genuinely novel. For instance, Elon Musk marveling at the complexity of hands overlooks that artists, surgeons, musicians, and neuroscientists have long understood this. Similarly, Palmer Luckey claimed no postmortem existed for the One Laptop Per Child project, ignoring the book The Charisma Machine. The absurdity culminates in Juicero, a $400 juicer that merely squeezed juice packs by hand. Discovering something new to oneself is exciting, but assuming it is new to everyone reflects a lack of intellectual humility.
The Professional Impact
This hubris has seeped into Silicon Valley's professional culture. Previously, technology aimed to serve customer needs, but post-financial crisis, entrepreneurs began inventing futures for consumers to accept. This mimics Steve Jobs' early failures before he returned to Apple with products like the iMac, iPod, and iPhone, which solved real problems. Today, companies chase trends like NFTs, the metaverse, and LLMs without addressing market needs. These innovations primarily benefit VCs and companies, not consumers.
AI has some utility, such as organizing data, but its widespread consumer adoption depends on being free. The only viable customer for LLMs may be the US government, but few will win contracts. OpenAI's Sam Altman suggesting ChatGPT for parenting advice ignores that children survived for decades without AI, thanks to sanitation, vaccines, and antibiotics. Similarly, Elon Musk's humanoid robot servants overlook existing labor-saving devices like dishwashers and dryers, which offer value without AI.
Consumer Disconnect
Normal people do not seek to automate every aspect of life. Vacation planning, for example, is a pleasure that builds anticipation. LLMs may assist in data organization or coding, but most people have no interest in coding. The primary consumer use for LLMs is cheating on schoolwork or replacing Google Search, which has degraded over time. However, LLMs' inaccuracies and plagiarism threaten the websites that generate quality information.
Inefficiency can be beneficial, as seen in stock market trading halts that prevent panic. In contrast, crypto's 24/7 trading exacerbates crashes. AI music apps assume people want to create music without learning instruments, but most enjoy listening. AI-generated content primarily benefits scammers, making it harder for genuine artists to earn a living.
The Root Cause
The disconnect arises because tech elites lack understanding of normal people's lives. They are influenced by VC podcasts and hype cycles, leading to products like NFTs and VR headsets that appeal only to a narrow group. Marc Andreessen's interview highlights this: he values company success over personal happiness, unlike most people who prioritize well-being. The real future lies in giving people what they actually want, not dictating trends.
