A leading artificial intelligence researcher has warned that AI development is accelerating beyond traditional exponential growth, entering what some experts describe as a “compounding exponential” phase — where advances build on each other faster than ever before.
Jack Clark, co-founder of Anthropic and former policy lead at OpenAI, first described this shift as early as 2023. He argued that AI progress followed three distinct phases:
- 1960s–2010: Mostly linear progress
- 2010–2020: Clear exponential growth driven by deep learning
- 2021 onward: “Compounding exponential” acceleration, where AI systems help accelerate their own development
“In other words, the next few years will yield progress that intuitively feels unbelievable,” Clark said.
Massive Advances Since 2023 Confirm Acceleration
Since Clark’s original comments, AI progress has rapidly intensified. Modern systems now demonstrate capabilities that were considered years away just a short time ago.
Major breakthroughs include:
- AI models capable of advanced reasoning, scientific analysis, and software engineering
- Systems that can generate realistic video, images, voice, and full applications
- AI agents that can autonomously complete complex multi-step tasks
- Models capable of assisting in scientific discovery, including drug research and materials science
Anthropic’s Claude models, including Claude 3 and newer versions, now rival or exceed human performance on many professional benchmarks, including legal analysis, programming, and graduate-level exams.
Meanwhile, OpenAI’s latest models — including successors to GPT-4 — show major improvements in reasoning, reliability, and autonomy, moving closer to what researchers call general-purpose AI systems.
AI Is Now Delivering Massive Economic Impact
Unlike earlier AI waves, today’s systems are already generating measurable economic value.
AI is now widely used in:
- Software development
- Customer support automation
- Scientific research
- Finance and legal analysis
- Media and content production
Major companies including Microsoft, Google, Amazon, and Meta are investing tens of billions of dollars into AI infrastructure, signaling expectations of transformative economic impact.
Clark previously warned that society had only a “tiny sliver of time” to prepare for rapid AI advancement — a concern now echoed by many leaders in the field.
Claude and ChatGPT Are Now Direct Competitors
Anthropic’s Claude chatbot has evolved into one of the strongest competitors to ChatGPT.
Claude models are known for:
- Strong reasoning and analysis abilities
- More cautious and safety-focused responses
- Excellent long-document understanding
ChatGPT, developed by OpenAI, remains one of the most widely used AI systems globally, with hundreds of millions of users.
Competition between companies like Anthropic, OpenAI, and Google has dramatically accelerated AI progress.
Even AI Leaders Warn About Limitations
Despite rapid improvements, AI still has limitations.
Sam Altman, CEO of OpenAI, has repeatedly cautioned that AI systems can still:
- Make mistakes
- Produce incorrect or misleading information
- Lack true understanding or consciousness
Altman has emphasized that current AI is powerful but not yet fully reliable or autonomous.
The Next Few Years Could Be Transformational
Many researchers now believe the coming years may represent one of the fastest technological revolutions in human history.
AI systems are increasingly able to:
- Improve software and technology development
- Assist scientific breakthroughs
- Enhance productivity across nearly every industry
Some experts believe this “compounding exponential” phase could eventually lead to artificial general intelligence (AGI) — systems capable of performing most intellectual tasks humans can do.
While timelines remain uncertain, one thing is clear: AI is advancing faster than almost anyone predicted just a few years ago.