LLMs, unlike prior transformative tech like the internet, disproportionately benefit regular people, with a muted, slower impact on corporations and governments
> They had to permit personal devices to be used work, but wanted to see enough justification to do that = catch-22. Took a while before an employee could use gmail, calendar, slack, Luke Maj / @lukemaj_ : @karpathy Good point! AI enables regular people to leverage expert capabilities and create a ‘1 person team’. Now with LLM help people can turn themselves into 1 person company. If a user is able to maximize efficient working with AI and agents he can produce far more output than an expert @jamesdouma : @karpathy Many of these traits also apply to internet search - it adds cheap and accessible broad+shallow capabilities to the general population. This empowers individuals more than organizations by the mechanisms you describe. Nick Dobos / @nickadobos : @karpathy Interesting point Ai might not beat subject matter experts. But it sure beats mediocre or busy coworkers, or not having any of those resources/training available Organizations benefit less because they go wide by hiring lots of people Meanwhile a person benefits hugely from Akshay Krishnaswamy / @hyperindexed : “...organizations deal with problems of a lot greater complexity...various integrations, legacy systems, stringent security protocols, privacy considerations, regulatory compliance and legal risk...” Organizations need the Ontology [image] Dan Mac / @daniel_mac8 : 🎥 reactions to “Power to the people: How LLMs flip the script on technology diffusion” from @karpathy tldr; AI is the first egalitarian general purpose technology 👇 [video] @jamesdouma : A nice framework for understanding the impact of LLMs: for now the overall impact is more about making everyone pretty good at everything rather than making experts more expert. The latter is important long term but the former happens sooner. (Like Google search) @heyskylark : who knew the discovery of general intelligence would happen in such an equitable way I think we're blessed to be living in a timeline of AI discovery that has played out this way so far
Context & Ripple Effects
Earlier coverage framed chat-based LLMs as a general-purpose information-processing interface and highlighted their potential to raise the output of less-skilled workers. This argument adds a diffusion lens: individuals can adopt these capabilities without the integration and governance burdens that slow institutions.
The contrast matters because prior expectations of a software-creation boom centered on lower barriers to making products; the case for radically easier software creation now extends to how much work a single person can plausibly coordinate.
First-order effects
- Individuals can use LLMs to draft, research, code, and organize work immediately, increasing the practical scope of a one-person team.
- Corporations and governments face a slower path to equivalent gains because deployment must fit legacy systems, security, privacy, compliance, and legal controls.
Second-order effects
- The near-term advantage shifts toward workers and small operators that can redesign their own workflows quickly, while larger organizations must justify and standardize use cases before broad rollout.
- Enterprise AI vendors face pressure to turn broad model capability into governed integrations and workflow products; raw access alone does not remove institutional adoption friction.
Third-order effects
- If this pattern persists, AI's earliest economic effects may be more decentralized than earlier enterprise-led technology waves, with individual productivity improving before institutional productivity statistics do.
- The eventual differentiator may shift from model access to organizational redesign: institutions that resolve governance and integration constraints could convert broad individual experimentation into durable scale.
The trend: LLMs are emerging as a consumer-first general-purpose technology whose broad access can diffuse capability faster among individuals than within regulated, complex organizations.