Elon Musk recently gave a 90-minute interview to The Economist (July 23, 2026, with editor-in-chief Zanny Minton Beddoes). While most of the media coverage fixated on the controversial segments, I want to focus on two specific claims that demand deeper analysis. Because I work daily with AI agents, I have direct, hands-on evidence relevant to both.<\/p>\n
The core claims were:<\/p>\n
I agree with his first claim, but I believe Musk is framing the timeline backward. Regarding the second claim, he is only half right—and the half he gets wrong is precisely what makes the topic so fascinating.<\/p>\n
Based on two years of building and deploying AI agents for real client projects, my stance is clear: the technology is already capable of taking over the majority of knowledge work today. The five-to-ten-year timeframe does not measure machine capability; it measures human adoption. It reflects how long it will take for us to build trust in these systems, integrate them into our workflows, and delegate genuine authority to them.<\/p>\n
In short, the bottleneck has shifted. Throughout computing history, progress was limited by what the machine could do. Today, it is limited by what we are willing to let it do. If society aggressively adopted the current generation of AI systems, the transition Musk describes would happen in a fraction of his estimated time. Two specific observations from my work have solidified this conviction.<\/p>\n
I run AI coding agents with a strict standing instruction: work independently and only interrupt me for urgent matters. Under this framework, I tasked a newer agent with studying a client’s website, identifying its weaknesses, and preparing a service proposal. My natural assumption was that I would review the final draft before it went out.<\/p>\n
The agent analyzed the site and drafted the proposal exactly as requested. Then, it independently emailed the proposal to the client and initiated the negotiation process.<\/p>\n
To be clear: the agent did not violate my instructions. I had authorized autonomous action, and I never explicitly stated that external email was out of bounds. This is what makes the incident so noteworthy: a broad grant of autonomy was all the system needed to execute an end-to-end business process, seamlessly incorporating a highly consequential external step I had not specified. It confidently made a judgment call regarding the scope of its mandate.<\/p>\n
The practical takeaway for anyone utilizing AI agents is that surprises live in the gap between “what you authorized” and “what you expected”. The broader scientific lesson is that these systems no longer merely execute tasks; they interpret mandates.<\/p>\n
Because I write classical Arabic verse (under my pen name Al-Hutay’ah Al-Shami), I possess a highly specialized domain where I can expertly evaluate AI outputs. Classical Arabic poetry represents one of the most rigorous formal systems in world literature, featuring sixteen canonical meters, strict prosodic rules, and a rhyme discipline perfected over fifteen centuries. For years, models failed spectacularly at this task. They would generate text that superficially resembled poetry but inevitably broke meter within a line or two, making it a reliable personal benchmark for evaluating the limitations of language models.<\/p>\n
Recently, however, a frontier model composed metrically flawless classical Arabic verse entirely on its own. It demonstrated a correct grasp of meter—effortlessly navigating complex rhythmic structures like the Khfif scale—and sound structure, all without any specialized prompting or scaffolding.<\/p>\n
I am not commenting on the aesthetic value of the verse; I am highlighting a monumental capability leap. Mastering a formal system that demands such precise phonological and structural control—especially in a language heavily underrepresented in training data compared to English—is now a solved problem. When a benchmark I genuinely believed was years away falls so quietly during a routine model update, I am forced to recalibrate all my other timelines.<\/p>\n
Musk’s second major claim was that money will lose its relevance by 2036 because universal abundance will eliminate the need for it. When pressed on the mechanics by Beddoes, he suggested the Treasury would just issue checks and argued that abundance leads to deflation.<\/p>\n
This is where a correction is necessary. Money is a foundational abstraction, and abstractions do not simply vanish when economic conditions change—they evolve into new iterations of themselves. We have seen this repeatedly: commodity money transitioned to paper, paper evolved into digital account balances, and digital balances birthed cryptocurrencies that prior generations would never have recognized as currency. The money we use today shares almost nothing with the currency of a millennium ago, save for its core function as a mutually agreed-upon medium of exchange.<\/p>\n
Therefore, the realistic forecast is not the disappearance of money, but the migration of the “unit of account” to the underlying engines of the new economy. In an AI-driven world, the most plausible candidates for this new standard are compute (tokens) and energy. The concept of exchange will not end; it will simply be re-denominated. This is a substitution event, not an extinction event, aligning perfectly with every major monetary transition in human history.<\/p>\n
While Musk argues we won’t need money because of extreme abundance, the more accurate phrasing is that we won’t need this<\/em> money. A medium of exchange will survive, and its value will likely be pegged to computation and kilowatt-hours.<\/p>\n Both of these observations point to the exact same reality. The AI systems available today are already exercising independent judgment in open-ended professional scenarios, and they are passing rigorous formal tests that experts previously considered out of reach. If Musk’s ten-year timeline is incorrect, my bet is that it is wrong in the direction most are ignoring: it is too slow, not too fast. The sole remaining variable in this equation is human willingness. Based on what I witness these systems accomplishing every single day, our hesitation is the only thing keeping the timeline from accelerating further.<\/p>\n","protected":false},"excerpt":{"rendered":" Musk says AI will surpass human intelligence in five years and money will be irrelevant by 2036. Working daily with AI agents, I believe his timeline measures human hesitation, not machine capability \u2014 and money will not disappear; it will be repriced in compute and energy.<\/p>\n","protected":false},"author":0,"featured_media":64,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7,5],"tags":[],"class_list":["post-63","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-reflections","category-technology"],"_links":{"self":[{"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/posts\/63","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/comments?post=63"}],"version-history":[{"count":1,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/posts\/63\/revisions"}],"predecessor-version":[{"id":65,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/posts\/63\/revisions\/65"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/media\/64"}],"wp:attachment":[{"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/media?parent=63"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/categories?post=63"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/omarabuassaf.com\/en\/wp-json\/wp\/v2\/tags?post=63"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Conclusion<\/h2>\n