AGI or Not, the AI Revolution is Already Here
.jpeg)
Artificial General Intelligence (AGI)—AI systems capable of performing any intellectual task a human can—has long dominated speculative discussions. Recent developments indicate AGI might be nearer than previously thought. In a widely-discussed episode of The Ezra Klein Show, Ben Buchanan, former special adviser for artificial intelligence in the Biden administration, revealed that the U.S. government is actively preparing for AGI, suggesting it could arrive within the next few years, potentially even within the current presidential term. Leaders in AI broadly agree, even if they're partly motivated by justifying current valuations: in recent interviews, Dario Amodei (Anthropic CEO), who prefers the term "powerful AI" to AGI, believes it will arrive by 2026 or 2027, and Demis Hassabis (Google DeepMind CEO) predicts AGI emerging within 5–10 years.
At the same time, there are still a number of AI experts and researchers who don’t believe AGI is possible with the current transformer-based paradigm for LLMs, given the inherent technical limitations of this approach.
But whether AGI arrives tomorrow or in a decade, today's AI has already changed everything. Rather than speculating endlessly about AGI, we should recognize—and urgently respond to—the AI revolution happening right now.
Today’s top large language models (LLMs)—such as GPT-4.5 from OpenAI, Claude 3.7 from Anthropic, and Google's Gemini 2.0 Pro at the time of this writing—have already become highly effective at handling complex tasks. Those making fun of AI’s occasional failures or limitations are missing the point entirely. These models don't merely promise future capabilities; they're powering transformative applications right now.
OpenAI’s recent release of its agentic "Responses API," for example, allows developers to create autonomous AI agents capable of independently handling complex workflows, from customer service interactions and content generation to complex planning tasks.
Meanwhile, China's startup Monica recently launched Manus, an autonomous agent that independently handles intricate responsibilities like sorting resumes, analyzing stock market trends, and even creating entire websites. Although Manus has provoked both enthusiasm and skepticism—often compared to China's advanced DeepSeek model—it undeniably showcases the level of automation and intelligence already achievable.
The critical takeaway is clear: even if all LLM research stopped today, current technologies already enable significant economic and societal transformations. The AI ecosystem has matured enough to produce powerful, tangible applications that will reshape our lives faster than most people anticipate.
Popular discussions about AGI typically assume some sudden, dramatic technological breakthrough. In reality, the building blocks for AGI largely already exist. Achieving AGI may be less about revolutionary breakthroughs and more about evolutionary refinements and integrations of existing components.
Anthropic’s recent Model Context Protocol (MCP), for example, enables AI assistants to seamlessly connect to diverse data sources and intelligently leverage context—dramatically improving their real-world usefulness. These incremental advancements—powerful autonomous agents, robust large language models, and increasingly seamless data integrations—collectively form the connective tissue for the emergence of general intelligence.
In practical terms, this means the LLMs we have today remain massively underutilized. Most businesses and individuals have barely begun scratching the surface when it comes to leveraging current AI technologies in daily operations.
As AI continues its rapid advance, its impact on our economy and society will be significant and immediate. Many existing roles—particularly repetitive or structured functions in sectors like customer support, finance, healthcare administration, and education—are already being automated. For example, AI-driven customer support agents have already reduced average response times by more than half at some leading companies, rapidly reshaping the customer service landscape and redefining job roles.
At the same time, new job categories we can't fully predict yet will inevitably emerge. History has repeatedly shown how technological revolutions—the internet or mobile computing—reshape job markets faster and more dramatically than most people foresee. The AI revolution will be no different, likely faster and even more disruptive.
Of course, rapid AI adoption also brings significant challenges—from ethical dilemmas and data privacy issues to managing massive workforce displacement. These challenges deserve careful consideration and thoughtful policy responses alongside continued technological advancement.
Given this reality, proactively adapting is the only rational strategy. Complacency or resistance will inevitably lead to obsolescence. Individuals, companies, and policymakers must begin actively engaging with and understanding AI to remain relevant and competitive.
Effectively learning to collaborate with AI might not guarantee security in your current role, but it significantly increases your chances of successfully transitioning into new, emerging roles. Foundational skills around AI literacy, effective prompting, and AI ethics & risk management are already becoming crucial.
What we already see people starting to figure out is that the first iteration of your role will be the AI-augmented version of it. That means: you learn to use AI tools beyond the basics and drastically increase your output (without losing quality, potentially increasing it).
But as more tasks become automated and AI systems handle more of the 'doing,' many new job opportunities will likely revolve around two of the most important aspects for any LLM-based use case: context management and output evaluation.
Context management refers to the crucial process of ensuring that AI systems operate with the right inputs—accurate, timely, and clearly structured data—so that they consistently produce useful and relevant outputs. As AI increasingly handles complex tasks, context management roles will involve curating, updating, and refining the datasets and knowledge sources that AI models rely on. Output evaluation, on the other hand, involves critically assessing the quality, accuracy, reliability, and ethical implications of AI-generated results. Professionals in these roles will be tasked with fine-tuning AI performance, identifying biases or inaccuracies, ensuring regulatory compliance, and continuously improving AI systems to align closely with organizational goals and societal values. Together, these two functions represent the essential "human-in-the-loop" oversight needed to responsibly harness AI’s transformative potential.
Preparing for these opportunities today means proactively building expertise in prompt engineering, data management, AI interpretability, and ethics—skills likely to be foundational for future roles.
Endless debates about who will "win" the AI arms race or precisely when AGI will appear distract from the immediate, practical reality: the AI revolution has already begun. Its transformative power is undeniable, and the tools for significant change are already here.
Society must urgently shift its attention away from speculation about distant future scenarios and towards tangible, immediate preparedness. Engaging with AI is no longer optional—it's essential. The best moment to start preparing was yesterday; the second-best moment is right now.
